The search engine’s definitions have been replaced with AI Overviews.
Google appears to be running into some hiccups after the company began rolling out its updated, and even more AI-focused search experience at I/O 2026. Currently, searching for the words “disregard,” “stop” or “ignore” on Google no longer displays a snippet with a definition, and instead offers an AI Overview and a lot of blank space. Because users have complained about the issue on social media, and publications like TechCrunch and Macrumors have reported on it, even if you don’t get a definition, you might still get a collection of links to articles documenting the issue before the traditional list of links.
Multiple members of Engadget’s staff were able to recreate the strange AI Overview responses with their own personal Google searches. In Incognito Mode, Google responded correctly once by displaying its usual snippet with the definition, and failed a second time by once again responding with an AI Overview. Links to online dictionaries still appear under these incorrect results, but you have to scroll past an AI Overview or a grid of articles to actually get to them.
Ian Carlos Campbell for Engadget
Engadget has contacted Google for more information about this issue and its attempts to fix it. We’ll update this article if we hear back .
In the grand scheme of things, Google not automatically displaying a definition isn’t as bad as recommending people put glue on pizza, one of the issues the company dealt with when it first launched AI Overviews. It might even be good for Merriam-Webster’s web traffic. What the issue does highlight is the awkward transition Google is currently undergoing, as it moves from the ultimate referrer of other websites into all-in-one AI assistant.
When agentic workflows fail, developers often assume the problem lies in the underlying model’s reasoning abilities. In reality, the limited information provided by the retrieval interface is often the primary limiting factor.
Researchers at multiple universities propose a technique called direct corpus interaction (DCI) that lets agents bypass embedding models entirely, searching raw corpora directly using standard command-line tools.
In classic retrieval systems such as RAG, documents are chunked, converted into vector representations (or embeddings), and indexed offline in a vector database. When an AI system processes a query, a retriever filters the entire database to return a ranked “top-k” list of document snippets that match the query. All evidence must pass through this scoring mechanism before any downstream reasoning occurs.
But modern agentic applications demand much more. “Dense retrieval is very useful for broad semantic recall, but when an agent has to solve a multi-step task, it often needs to search for exact strings, numbers, versions, error codes, file paths, or sparse combinations of clues,” the authors of the DCI paper said in comments provided to VentureBeat. “These long-tail details are precisely where semantic similarity can be brittle.”
Unlike static search, agents must also revise their search plans dynamically after observing partial or localized evidence. Exact lexical constraints and multi-step hypothesis refinement are difficult to execute with semantic retrievers. Because the retriever compresses access into a single step, any critical evidence filtered out by the similarity search cannot be recovered later, no matter how advanced the agent’s downstream reasoning capabilities are. As the authors explain, current retrieval pipelines can become a bottleneck because “they decide too early what the agent is allowed to see.”
This direct access addresses a core problem in enterprise environments: data staleness. Embedding indexes are always a snapshot of a specific moment in time, taking considerable compute and time to build and maintain.
“In many enterprise settings, the data is not a stable document collection. It is daily financial reports, live logs, tickets, code commits, configuration files, incident timelines, and internal documents that keep changing,” the authors said. DCI lets the agent reason over the current state of the workspace rather than yesterday’s vector index.
The agent operates in a terminal-like environment where its observations are raw tool outputs such as file paths, matched text spans, and surrounding lines. The core tools provided by DCI are few but highly expressive. Agents use commands like “find” and “glob” to navigate directory structures and locate files. For exact matching, they use “grep” and “rg” to locate specific keywords, regex patterns, and exact strings. When local inspection is needed, tools like “head,” “tail,” “sed,” “cat,” and lightweight Python scripts allow the agent to peek at the context surrounding a match or read specific file sections.
The agent can combine these tools via shell pipelines to execute complex search logic in a single step. An agent can pipe commands to enforce strict lexical constraints, such as searching a file for one term and piping the output to search for a second term. It can combine multiple weak clues across a corpus by finding a specific file type, searching for a keyword like “report,” and filtering for a year like “2024.” It can also immediately verify a hypothesis by inspecting the exact lines around a keyword match.
DCI delegates semantic interpretation directly to the agent instead of relying on embedding-based similarity search. The agent can formulate hypotheses, test exact lexical patterns, and extract detailed information that a traditional semantic retriever might miss.
The researchers propose two versions of this system. DCI-Agent-Lite is designed as a lightweight, low-cost setup built on the GPT-5.4 nano model and restricted purely to raw terminal interactions like bash commands and basic file reads. Because reading raw files can quickly fill up a smaller model’s memory, this version relies on lightweight runtime context-management strategies to sustain long-horizon exploration.
DCI-Agent-CC is the higher-performance version, designed for teams with more compute budget. It runs on Claude Code powered by Claude Sonnet 4.6. Claude Code provides stronger prompting, more robust tool orchestration, and superior built-in context handling, which improves the agent’s stability during complex, multi-step searches across heterogeneous datasets.
The researchers tested both versions of DCI across agentic search benchmarks like BrowseComp-Plus, knowledge-intensive QA with single-hop and multi-hop reasoning, and information retrieval ranking in tasks requiring domain-specific reasoning and scientific fact-checking.
They tested DCI against three baselines. The first included open-weight retrieval agents such as Search-R1 and proprietary agents powered by frontier models like GPT-5 and Claude Sonnet 4.6, paired with standard retrievers. The second baseline included classical sparse retrievers like BM25 and dense retrievers like OpenAI’s text-embedding-3-large and Qwen3-Embedding-8B. The third baseline consisted of high-performing reasoning-oriented re-rankers like ReasonRank-32B and Rank-R1.
DCI systematically outperformed the baselines, according to the researchers. On the complex BrowseComp-Plus benchmark, swapping a traditional Qwen3 semantic retriever for DCI on a Claude Sonnet 4.6 backbone improved accuracy from 69.0% to 80.0% while reducing the API cost from $1,440 to $1,016. The return on investment for lightweight agents was also noticeable. DCI-Agent-Lite with GPT-5.4 nano competed with the OpenAI o3 model using traditional retrieval while cutting costs by more than $600.
On multi-hop QA benchmarks, DCI-Agent-CC reached an 83.0% average accuracy, improving on the strongest open-weight retrieval baseline by 30.7 points, according to the researchers.
The data shows that DCI has lower overall document recall than dense embedding models, but once it finds a relevant document, it extracts substantially more value from it.
“If an enterprise AI lead asked where DCI is most clearly useful, I would point to tasks that require exact evidence localization in a dynamic workspace: debugging production incidents, searching large codebases, analyzing logs, compliance investigation, audit trails, or multi-document root-cause analysis,” the researchers note.
In one complex deep-research task, the agent had to identify a specific soccer match based on 12 interlocking clues, including exact attendance, yellow cards, and player birth dates. A traditional retriever would fail by surfacing short, disconnected snippets. Instead, the DCI agent explored the file directory, read specific lines of a 1990 England versus Belgium match report to verify the exact number of substitutions, pulled a specific quote from an interview file, and verified the exact birth dates of two players by peeking into their Wikipedia text files. By chaining these simple commands, DCI ensures that no evidence is permanently lost behind a flawed semantic search algorithm.
DCI has a clear operating envelope where it scales excellently in search depth but struggles with search breadth. When the experimental corpus was expanded from 100,000 to 400,000 documents, the system’s accuracy dropped significantly and the average number of tool calls rose. While DCI is powerful once a promising document is found, the cost of locating that initial useful anchor document grows sharply as the size of the candidate space increases.
DCI also has lower broad document recall compared to dense embedding models. It trades exhaustive recall for high-resolution, local precision. If an enterprise workflow strictly requires finding every single relevant document across a massive dataset, DCI may not be the right tool.
Granting an agent expressive tools like an unrestricted bash shell increases latency and compute costs due to the high volume of iterative tool calls required to complete a search. It also creates significant context-management and security challenges for IT departments.
“Tool calls can return large outputs; long trajectories can fill the context window; and raw terminal access requires sandboxing, permission control, and careful engineering,” the authors said. To manage the context window, the researchers found that moderate truncation and compaction help the agent sustain longer searches, whereas overly aggressive summarization tends to discard useful evidence.
Because of these operational realities, DCI is not meant to be a mandatory replacement for existing vector infrastructure. Instead, it serves as a complementary one.
“For orchestration engineers and data architects, our view is that the most practical near-term deployment pattern is hybrid,” the authors said. Semantic retrieval can still provide high-recall candidate discovery when a user’s intent is broad or underspecified. “DCI can then operate as a precision and verification layer: the agent can search within the retrieved documents, expand from them into neighboring files, check exact constraints, and combine weak signals across documents.”
The researchers have released the code for DCI under the permissive MIT license.
“Longer term, DCI changes how we think about enterprise data. Data will not only need to be stored for humans or indexed for search engines; it will need to be organized for agents that can inspect, compare, grep, trace, and verify,” the authors conclude. “File names, timestamps, stable identifiers, metadata, version history, and machine-readable structure become part of the retrieval interface.”
The new standalone app sits on top of Facebook Groups, adds an AI tab called “Ask” and an admin assistant, and arrives the same month Mark Zuckerberg told staff he and Chris Cox had discussed whether they could build 50 new apps.
Meta has released a new standalone app called Forum, without a launch event, a blog post, or much of a press push at all.
The product is built on top of Facebook Groups and is listed in the App Store as “a dedicated space built for deeper discussions, real answers and communities you care about,” phrasing that does most of the work of telling you what Meta wants it to be: a Reddit, with a Facebook account attached.
The launch was first spotted by the social-media consultant Matt Navarra and reported by TechCrunch, with independent confirmation from Engadget and MacRumors. A Meta spokesperson told Engadget the product was still in testing:
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
“We test lots of new products publicly to see what people find interesting and useful to their experiences across our apps.”
Users sign in with their existing Facebook account, at which point their groups, profile and activity carry over. Posts can be made under a nickname, in the same way they can in the main Facebook app, although group administrators can still see real identities.
Anything posted on Forum shows up in the corresponding group on Facebook, and vice versa, so the product is less a separate network than a separate front door into the one Meta already has.
The feed is the differentiator. Where Facebook’s main timeline mixes friends, Pages, algorithmic suggestions and ads, Forum’s surfaces only conversations from groups a user is in, with prompts to discover others. Two AI features sit on top.
The first, “Ask,” lets users put a question to the app and receive an answer compiled from discussions across groups, sparing them the chore of searching one at a time.
The second is an admin assistant designed to help moderators run groups and handle moderation. Both are pitched as time-savers rather than as the product’s defining feature.
This is not Meta’s first standalone Groups app. The company shipped one in November 2014 and shut it down in 2017, for reasons it never made especially clear at the time.
The Reddit framing is also conspicuous given timing: Reddit went public in March 2024, has spent the past two years licensing its data to AI companies for training, and remains the closest thing the consumer internet has to a working community-discussion product at scale. A Facebook-flavoured competitor with an AI answer tab is, at minimum, a recognisable shape.
Forum is the second new Meta app in roughly a month. In late April, the company began testing Instants, a standalone Instagram companion for disappearing photos that borrows visibly from BeReal and Snapchat. Meta Edits, last year’s CapCut-shaped video editor, sits in roughly the same lineage.
That cadence is not an accident. The Wall Street Journal reported earlier this month that Zuckerberg told staff in an internal Q&A that AI-driven efficiency was allowing Meta to build more products with smaller teams, and that he and chief product officer Chris Cox had discussed whether the company could ship 50 new apps.
He then talked himself partway back down: “Like, yeah probably. But we probably should start by doing a few before we just, like, ramp up trying to do 50 all at once.”
Forum is one of the few. Whether users want a separate app for their Facebook groups, with an AI tab attached, is the question the spokesperson’s testing line was designed not to answer. The 2014 Groups app suggests one historical data point. The fate of Instants, Edits, and whatever ships next will produce the rest.
A navigation issue affecting the My Pixel app has left a number of Pixel owners unable to access key sections of the preinstalled Google app, with the bottom tab bar disappearing entirely from the interface without warning.
The missing bar ordinarily carries four tabs covering Home, Tips, Support, and Store, giving users a single access point for feature guidance, troubleshooting help, and direct links to Google’s hardware shop from within the device itself.
Reports from affected users on Reddit confirm the tabs have vanished from their devices, with at least one account suggesting the issue followed a dismissal of a pop-up notification that appeared when the user attempted to navigate to the Google Store section of the app.
That sequence points toward a possible server-side or account-level trigger rather than a local software fault, though Google has not acknowledged the issue publicly or provided any indication of whether a fix is in progress.
The timing is notable given that My Pixel serves as one of the primary in-app channels through which Google surfaces Pixel-specific feature tips and direct hardware purchasing options, making a navigation failure more disruptive than a cosmetic glitch would suggest.
For users who rely on the app’s support tab as their first point of contact when troubleshooting device issues, the disappearance of the tab bar effectively removes that route entirely until the problem is resolved, leaving the Google Support website as the most direct alternative in the interim.
Google has not confirmed whether the issue affects specific Pixel models, specific Android versions, or a broader cross-section of the Pixel install base, and the scope of the problem remains unclear based on available reports.
That silence extends to any fix or workaround, with Google yet to set out a timeline for restoring full navigation functionality to affected devices or clarify whether the issue traces back to a server-side change, an app update, or something else entirely.
The Texas Attorney General has sued Meta over allegations that the company’s WhatsApp messenger, used by more than 3 billion people, doesn’t provide the end-to-end encryption (E2EE) it has long claimed.
Since at least 2016, Meta (then named Facebook) has said WhatsApp provides robust end-to-end encryption, meaning that messages are encrypted on a sender’s device with keys that are available only to the receiver’s. By definition, E2EE means that no one else—including the platform itself—can read the plaintext messages.
In sworn testimony before two US Senate committees in 2018, CEO Mark Zuckerberg said Meta does “not see any of the content in WhatsApp; it is fully encrypted” and that “Facebook systems do not see the content of messages being transferred over WhatsApp.” The engine for this E2EE is the Signal protocol, an open source code base that multiple third-party experts have said lives up to its promises.
In a complaint filed Thursday, Texas AG attorneys said Meta’s claims are false and that the company can and does read the unencrypted contents of WhatsApp messages. They said they are filing the action to “prevent WhatsApp and Meta from continuing to willfully deceive [Texans] by misrepresenting that their private communications were just that—private and inaccessible even to WhatsApp and Meta—when, in fact, WhatsApp and Meta have access to all WhatsApp users’ communications in their entirety.”
“The gravity of Meta’s and WhatsApp’s violation of users’ privacy and trust cannot be overstated,” the attorneys wrote. “All users were entitled to believe their communications were private when WhatsApp and Meta unequivocally and repeatedly promised that no one—not even WhatsApp and Meta—can access their messages.”
In an email, Meta called the allegations “baseless” and vowed to fight the lawsuit in court.
The sole factual evidence cited for the claims is an article published last month by Bloomberg. It reported that the US Commerce Department’s Bureau of Industry and Security had abruptly closed an investigation into allegations that Meta could access encrypted WhatsApp messages shortly after one of the department’s agents sent an email outlining the probe’s preliminary findings.
Most GPS watches make you choose between rugged durability and genuinely useful health tracking, which is exactly why the Garmin Instinct 3 Solar stands out.
The Garmin Instinct 3 Solar makes no such compromise, and it is currently down from £299.99 to £209.99 at Argos, saving you £90 in the process.
A 30% drop puts the Garmin Instinct 3 Solar firmly in “upgrade‑worthy” territory for your fitness tracking
At 30% off, the Garmin Instinct 3 Solar is a serious offer for anyone who trains hard, travels far, and loves a great deal.
The solar charging lens is the detail that separates this watch from the rest of the field, because it means battery anxiety simply stops being part of the conversation on longer adventures and multi-day expeditions.
With up to 28 days in typical smartwatch mode and up to 40 days in low-usage mode, you are dealing with a device designed to outlast your plans rather than dictate them from a low-battery warning.
That endurance would mean little without the hardware to back it up, and the Garmin Instinct 3 Solar delivers a metal-reinforced bezel, scratch-resistant lens, and a construction built to survive the kind of conditions that would retire a lesser watch permanently.
The health tracking suite is equally serious, covering wrist-based heart rate, advanced sleep monitoring, Pulse Ox, fall and crash detection, and activity tracking across running, cycling, swimming, and hiking, all feeding back through Garmin’s own operating system.
NFC connectivity means you can also use the watch for contactless payments, which matters on runs or rides where carrying a phone or wallet isn’t an option.
The built-in LED flashlight with variable intensities rounds things out in practical terms, giving you a genuinely useful tool rather than a novelty feature when visibility drops during early-morning or evening sessions.
At 30% off, the Garmin Instinct 3 Solar is a serious offer for anyone who trains hard, travels far, or simply refuses to babysit a charger every other night just to keep their wrist tech alive.
SQUIRREL_PLAYLIST_10148964
I don’t know how much of my life has been spent hovering over a grill, stove, smoker or oven trying to figure out if the thing I’m cooking for family and friends is actually ready, but I can tell you it’s more than it needed to be. Last year, I picked up one of these Chef iQ wireless probes for a big dinner I was hosting, and now I pull these things out to use them at least once a week. Today, you can get them for way less than I did.
This kit is a pair of Chef iQ probes in a charging case, which, in my experience, will fully charge both probes twice before it needs charging itself. You pop the probes into whatever you’re cooking, and you can either check the temperature in the app on your phone or set an alert when the thing you’re cooking reaches its desired temp. No opening the oven or smoker to let all the heat out, leading to a more efficient and balanced cook. These genuinely improved my cooking, and now I’ve given them to my brothers to improve theirs.
Memorial Day just dropped this set to $100, and I’ve only ever seen it cheaper once before. Get yourself one of these, or give them as a gift, and we can all end the act of hovering over the grill.
Salesforce has closed 29,000 Agentforce deals and reports $800 million in ARR, but its stock is down 30 per cent in 2026 amid the SaaSpocalypse selloff. Showcase demos from Williams-Sonoma, UChicago Medicine, and SharkNinja turned out to be works in progress rather than live deployments.
TL;DR
Salesforce has a problem that no amount of marketing can fix. The company has built its entire narrative around Agentforce, its AI agent platform, and the numbers look impressive on paper: 29,000 deals closed, $800 million in annual recurring revenue, and a roadmap that promises to replace entire categories of human work. But Wall Street is not buying it, and the gap between what Salesforce shows on stage and what customers actually use keeps getting wider.
The stock tells the story. Salesforce shares fell nearly 21 per cent in 2025 and have dropped another 30 per cent so far in 2026. The decline tracks a broader selloff in software-as-a-service companies, an event the market has taken to calling the SaaSpocalypse. Roughly $285 billion in SaaS market capitalisation evaporated in a single 48-hour window in February. The logic is simple: if one AI agent can do the work of ten employees, why would a company pay for ten seats?
Salesforce has tried to get ahead of that question by positioning itself as the company that sells the agents rather than the seats. CEO Marc Benioff has called Agentforce a “digital labour platform.” On earnings calls, the company cites the 29,000 deals and the ARR figure as proof that enterprises are buying in.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The trouble is that the showcase examples keep falling apart under scrutiny. At Dreamforce, Salesforce demonstrated a Williams-Sonoma AI agent called Olive that was supposed to act as an agentic sous chef, helping customers plan meals and find products. In practice, Olive struggled with specific questions and recommendations. The agent’s more advanced capabilities were described using future tense, “will soon be able to,” rather than as features that were live.
A similar pattern appeared with the University of Chicago Medicine. Salesforce presented the hospital system as a flagship Agentforce for Health deployment. The reality was more modest: UChicago Medicine’s first AI agent launched on web chat to handle basic questions like parking directions and clinic availability. The more ambitious features, including voice-based patient support, were still in development.
SharkNinja, the maker of Shark vacuums and Ninja kitchen appliances, was another headline customer. Salesforce said the company would use Agentforce to streamline customer service. Bloomberg reported a 20 per cent reduction in support calls as part of the pitch. But the deployment described was forward-looking, with agents expected to “guide customers through the buying process” and “manage returns,” not a report on outcomes already achieved.
This matters because Salesforce is not the only company overselling AI capabilities. Apple agreed to pay $250 million in May to settle a class action lawsuit alleging it had exaggerated what Apple Intelligence and a smarter Siri would deliver when it launched the iPhone 16. The settlement covered claims that the company’s marketing went well beyond what the technology could do at launch.
Salesforce’s financial trajectory adds another layer. Revenue growth has slowed from roughly 25 per cent a few years ago to about 10 per cent in fiscal 2026, when the company reported $41.5 billion in total revenue. That is still a large business, and the company delivered a strong fourth quarter with 12 per cent growth. But the deceleration is exactly what investors fear when they hear that AI agents will compress the number of human users who need software licences.
The company has tried to address the pricing question. Agentforce uses a consumption-based model rather than traditional per-seat pricing, charging for what Salesforce calls “agentic work units.” It has consumed nearly 20 trillion tokens and converted them into more than 2.4 billion such units. Whether that model can grow fast enough to offset the structural threat to seat-based revenue is the central bet.
Smaller customers illustrate both the promise and the cost. The city of Kyle, Texas, deployed Agentforce to run its 311 service, handling more than 12,000 resident requests since March 2025 with nearly 90 per cent first-call resolution. Bloomberg reported the city doubled its Salesforce spending to $300,000. For a fast-growing municipality, that may be a reasonable investment. For enterprise customers weighing the same calculus at scale, the economics are less clear.
The competitive pressure is real. SAP unveiled its Autonomous Enterprise with more than 200 AI agents and an Anthropic partnership at Sapphire 2026. ServiceNow, Google, and Microsoft are all building agent platforms. The question is no longer whether AI agents will reshape enterprise software but whether Salesforce can maintain its position as the market reprices around it.
Benioff has responded with characteristic confidence, announcing a new revenue target of $60 billion by fiscal 2030. He has also committed $50 billion in share buybacks, a signal to investors that the company believes its stock is undervalued. Slack’s transformation into an agentic platform, with more than 30 new AI capabilities and mandatory bundling with every new Salesforce account from this summer, is part of that push.
None of this resolves the core tension. Salesforce is asking customers to pay for a future that its own demos have not yet delivered, while asking investors to trust that consumption-based AI revenue will replace the seat-based model that built the company. The 29,000 deals are real. The $800 million in ARR is real. But the agentic AI market rewards outcomes, not announcements, and the gap between the two is where Salesforce’s credibility will be tested.
Presented by Veriff
Americans can’t reliably distinguish real from AI-generated content, and that’s not just a media literacy problem; it’s a direct threat to how businesses verify identity online.
New research finds that while many people are aware of deepfakes, their ability to distinguish them from reality is barely better than a coin flip. A 2026 survey conducted by Veriff and Kantar among 3,000 respondents in the United States, the United Kingdom, and Brazil shows Americans scoring just 0.07 on a scale where 0 represents random guessing.
If people can’t distinguish authentic visual content, they can’t reliably distinguish authentic identities. In practice, that means the same users interacting with digital services are often unable to tell whether the person on the other side of a screen is real.
That ineffectiveness has direct consequences for every digital business that relies on image- and video-based identity verification to confirm who is on the other side of a screen. That includes everything from customer bank onboarding and account recovery to marketplace seller verification, high-value ecommerce transactions, social platform authentication, and enterprise access control.
In the U.S., those consequences are already material — synthetic identity fraud now accounts for billions in annual losses, and the tools to generate convincing fakes are now widely accessible.
The report also identifies a small but high-risk cohort: the roughly 7% of users who perform poorly at detecting deepfakes, yet remain confident in their ability and rarely verify what they see. While this is small as a percentage, at scale it represents millions of accounts that are highly exploitable targets for fraud.
If users can’t reliably distinguish real from synthetic identities, then any system that depends on visual verification is fundamentally exposed. Identity verification can no longer be treated as a compliance function; instead, it has to be built as core digital infrastructure.
“Now that AI-generated content is becoming indistinguishable from reality, the human eye alone is no longer a reliable line of defense,” says Ira Bondar-Mucci, fraud platform lead at Veriff. “Businesses and policymakers in the U.S. need to close this awareness gap urgently, while simultaneously investing in automated verification technologies that can catch what humans simply can’t.”
The United States might be the global epicenter of generative AI development, but American consumers demonstrate the lowest familiarity with deepfakes among the three surveyed markets. Only 63% of U.S. adults are familiar with the term, compared to 74% in the UK and 67% in Brazil.
“There’s a paradox at play,” Bondar-Mucci says. “The U.S. is the global epicenter of AI development, yet American consumers are the least familiar with one of its most dangerous byproducts. Historically, consumers have had higher baseline trust in digital content, with the conversation about fraud centered more on data privacy than on content authenticity. The problem is that low awareness doesn’t reduce risk, it amplifies it. If you don’t know what a deepfake is, you’re far less likely to pause and verify whether you’ve encountered one.”
In practice, the randomness that characterizes consumer’s ability to distinguish real from fake is evident across the ways people assess different types of content. Video content proved to be especially difficult to assess, with fake videos frequently identified as authentic and real videos often flagged as fake. Even in side-by-side comparisons, respondents split their judgments close to evenly, another indication that visual inspection alone is no longer a reliable method for verifying authenticity.
Roughly half of U.S. respondents say they are confident in their ability to identify deepfakes, but that confidence far exceeds actual performance, demonstrating that self-assessment is effectively meaningless.
Within that population, there’s that small but high-risk cohort: the approximately 7% of users who are inaccurate, yet overconfident in their ability and rarely verify suspicious content.
“This confidence-competence gap creates a false sense of security that fraudsters are primed to use,” says Bondar-Mucci. “When people believe they can’t be fooled, they stop looking for the signs. That’s precisely when they’re most vulnerable, whether to a synthetic identity used in financial fraud or a fabricated video designed to manipulate trust.”
For businesses, the implication is clear: any organization that still relies on manual review processes or customer self-attestation is inheriting this vulnerability directly. Human judgment is an increasingly unreliable safeguard, and verification needs to be built into systems by default. This means automated, technology-led, and not dependent on the end user’s self-assessment of their ability to tell real from fake.
Concern about deepfakes is high across the U.S., with 79% of respondents reporting they are rather or extremely concerned about personal fraud and impersonation.
The U.S. diverges from other markets in where that concern gets directed. Americans are more likely than UK or Brazilian respondents to trust social media platforms and digital services to identify and manage AI-generated content. That delegation of responsibility may be reducing individual vigilance at exactly the moment the threat is accelerating.
“We’re seeing synthetic identities used to open fraudulent accounts and authorize transactions, and deepfake videos deployed to bypass basic verification checks,” he explains. “What makes this particularly urgent is the combination of great concern with relatively high platform trust. That gap between perceived and actual protection is exactly where fraud thrives.”
The gap between what Americans believe they can detect and what they actually can is not a knowledge problem that awareness campaigns will resolve, but a design flaw in any system that places the burden of identity verification on unassisted human judgment.
The effective response is not to remove humans from the verification loop, but to stop assigning them tasks that human perception can no longer perform reliably. Organizations that persist in relying on manual review processes or customer self-attestation are absorbing this vulnerability into their operations.
The alternative is automated, AI-powered identity verification that operates at the point of interaction, detects synthetic media before a human decision is required, and does not depend on the end user’s ability to distinguish real from fake.
“Seeing is no longer believing,” says Bondar-Mucci. “The companies that build verification infrastructure around that reality, rather than around the assumption that it will be otherwise, are the ones best positioned to sustain customer trust as the synthetic media landscape continues to evolve.”
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. For more information, contact sales@venturebeat.com.
Power grid operators and datacenter developers in the United States are in a bind, and energy analysts can’t see an easy way out.
The American power grid is old, outdated, and in desperate need of upgrades. Add in a growing number of gigawatt-scale AI datacenters demanding stable access to power that doesn’t disrupt service-level uptime agreements and things start to look even worse.
Energy costs have already skyrocketed in the nation’s largest energy market mainly thanks to the bit barn bonanza, leading to a new chorus of calls for datacenters to bring their own power generation if they want reliable supplies not bound by grid constraints.
According to energy analysts at Wood Mackenzie, datacenter operators have a choice to make, and neither option is great.
They can wait the five to 10 years it’ll take for grid operators to upgrade their transmission and generation capabilities to account for their demands.
Or they can accept deals with power companies that supply them with power but require curtailment during peak loads, and then install their own on-site power generation to make up the difference. This is the far riskier option, but it’s one that many operators are going with.
“With more than 90 GW of collocated generation in US interconnection pipelines, it is clear that the need to scale up at speed has sent many data centre developers down the riskier path,” WoodMac explained in the report. “Collocating volatile AI workloads with power generation has scarce precedent, though, and is far more difficult than most in the industry understand.”
Many datacenter operators are only thinking about generator megawatts, say the analysts, leaving engineers frustrated at having to explain the technical complexities of building colocated power generators. Multiple grid operators have passed rules that, as key stakeholders understand them, could give utilities priority rights over colocated power generators during shortages, potentially forcing datacenters to reduce demand while supplying power back to the grid, WoodMac said.
“This effectively makes the model unworkable,” the analysts said. “Few data centre developers would invest in baseload generation if it could not be utilised when it was most needed.”
Near-instantaneous swings in AI power demand can damage reciprocating engines and gas turbines, while batteries may not respond fast enough to every spike and can degrade over time. The rapidly fluctuating power demanded by hyperscale AI datacenters could even damage the grid itself.
“These loads can also cause sub-synchronous oscillations, which pose fundamental stability risk to not only local generators but also to distant ones on the transmission system,” the analysts note. “Technology providers are only beginning to come to terms with this challenge, the mitigation of which is site specific, making solutions hard to scale.”
In other words, on-site generation might solve one problem while introducing a whole host of new issues.
“While hyperscalers are likely to successfully operationalise some projects with collocated resources, it will come at considerable cost,” the report concludes. “This cost, along with the technical risk and project-specific mitigation required, will prevent collocation from emerging as a scalable model.”
So, what’s the alternative? Well, building out the grid, obviously. WoodMac said that grid operators largely don’t consider conditional interconnection with curtailment requirements to be a long-term solution, and are all investing billions in modernization. But those modernization initiatives are going to hit everyone in the wallet.
Improving grids to serve AI datacenters “has profound implications for affordability … regional network upgrades necessary to support local large-load connections will be spread among all ratepayers,” WoodMac said. “This may trigger a political outcry.”
Rates are already increasing across the US, said Ben Hertz-Shargel, WoodMac’s global head of grid transformation and large loads, and no presidential decree can slow that down.
“As the data center buildout eclipses existing utility capacity and more infrastructure must be built, we’ll see that increasingly become a driver of customer rates,” Hertz-Shargel told us in an email.
The study doesn’t present a solution for the current situation, forecasting uncertainty as grid operators hurry to come up with some solution, as datacenter operators’ plans blow past their ability to meet demand. Hertz-Shargel didn’t have much comfort to offer here, either.
“With utilities and grid operators reforming their load interconnection processes, we’re likely to see more capacity made available through utilities,” Hertz-Shargel said. Not everyone is going to be able to secure that utility capacity, though, and the organizations that are left out will be stuck with the same choice: Wait for the grid to catch up, or rely on colocated generation and conditional grid connection deals, Hertz-Shargel explained.
“When the current wave of transmission capacity completes, though, the industry will be in a position for quick acceleration,” Hertz-Shargel added – as long as the numbers continue to work out. “That presumes that investors like what they see in terms of AI’s return on investment by that time, however, and remain comfortable deploying massive capital into digital infrastructure.”
That’s far from a sure thing at this point.
The only thing that the report sees for certain in the current situation is a quick division into winners and losers in the AI race. Big players who are able to absorb the costs will emerge as victors.
“The challenges facing [energy] collocation are surmountable for the most experienced and deep-pocketed developers,” Wood Mackenzie said. “Companies capable of operating reliably without firm grid service will be able to scale their AI business faster than others, positioning them to outcompete.”
In other words, expect the big guns to further entrench their dominance as the little guys are starved to death or are gobbled up by the competition. There’s nothing like the American dream, eh? ®
Security
Will Jason Statham save us?
A malware-spreading scumbag swimming through GitHub pushed malicious commits to more than 5,500 repositories on Monday as part of an automated campaign called Megalodon.
Similar to the earlier TeamPCP attacks that poisoned about 3,800 GitHub repositories, this new campaign has so far infected 5,561 repos with CI/CD credential-stealing malware, according to SafeDep researchers, who uncovered the predatory commits and published a full list of the compromised repositories.
If a repository owner merges the commit, the malware executes inside their CI/CD pipeline and propagates further, Ox Security lead researcher Moshe Siman Tov Bustan said in a Thursday blog post.
Megalodon steals AWS secret keys and Google Cloud access tokens. It also queries AWS, Google Cloud Platform, and Azure metadata for instance role credentials, reads SSH private keys, Docker and Kubernetes configurations, Vault tokens, Terraform credentials, and scans source code for more than 30 secret regex patterns. Then it exfiltrates GitHub tokens, including secrets used to authenticate with cloud providers, thus allowing attackers to impersonate developers’ cloud identities, along with Bitbucket tokens.
In other words: consider ALL of your CI/CD variables pwned.
“We’ve entered a new supply chain attack era, and TeamPCP compromising GitHub was only the beginning,” Bustan told The Register. “What’s coming next is an endless wave, a tsunami of cyber attacks on developers worldwide.”
Plus, he added, hacking GitHub “compromises the security of every company with a private repository hosted on the platform.”
This new wave of supply chain attacks hitting developers’ environments won’t stop until “companies like npm and GitHub take serious action against the spread of malicious code on their servers,” Bustan said.
He noted npm’s statement on X saying it “invalidated npm granular access tokens with write access that bypass 2FA” to prevent additional supply-chain attacks like Mini Shai Hulud.
“That could help a little with account hijacking, but it doesn’t solve the actual problem,” Bustan said. “Malicious code is still reaching their servers, and nothing is stopping it before it does.”
SafeDep spotted Megalodon hidden inside a legitimate package: Tiledesk, an open source live chat and chatbot platform. The attacker backdoored versions 2.18.6 (May 19) through 2.18.12 (May 21), and the same npm maintainer published the last clean version, 2.18.5, before unknowingly publishing these newer compromised versions.
“The attacker never touched the npm account,” the open source supply-chain security startup researchers said. “They compromised the GitHub repository, and the maintainer published from the poisoned source without realizing it.”
While publishing malicious packages on npm is a TeamPCP signature move, Bustan said there’s no threat-intel or code-analysis evidence that connects Megalodon to the crew behind the Trivy, Checkmarx, and other recent supply-chain attacks. “Our best guess now is that it’s a different threat actor copying their behavior and style, but not much of the code itself,” he told us.
And despite TeamPCP open sourcing its Shai-Hulud worm and announcing a supply-chain attack competition on BreachForums, Ox doesn’t believe Megalodon is a contest entry.
“We have indications that they are not participating in the TeamPCP contest due to the contest having a specific rule to add a public encryption key that the actor behind the malware could match with his private key to prove his involvement,” Bustan said.
SafeDep’s threat hunters traced the malicious commit (acac5a9) to an author “build-bot,” connected to the email address build-system[@]noreply.dev with the message “ci: add build optimization step.”
The author name and noreply email mimic automated CI commits, and there’s no GitHub account linked to the author and committer user fields. “Someone pushed the commit to master with no PR and no merge commit, using a compromised PAT or deploy key,” according to the researchers.
They searched GitHub for other commits authored by the same email address and found 2,878 results, plus a second email, ci-bot@automated.dev, with an additional 2,841 commits. All landed May 18 during a six-hour window (11:36 to 17:48 UTC) and targeted 5,561 repositories.
This includes nine compromised Tiledesk repositories: tiledesk-server, tiledesk-dashboard, tiledesk-telegram-connector, tiledesk-llm, tiledesk-docker-proxy, tiledesk-community-app, tiledesk-campaign-dashboard, tiledesk-helpcenter-template, and tiledesk-ai. Others include Black-Iron-Project with eight compromised repos, WISE-Community, and hundreds of smaller repositories. ®
BloFin War of Whales 2026 Grand Prix opens registration for $5M trading championship
Google reimburses Register sources who were victims of API fraud
Napoleonic enters 2026 Doomben 10,000 field via Abounding withdrawal
BeInCrypto 100 Institutional Awards Nomination: KAST for Best Digital Assets Neobank and Best Digital Assets Fintech
Bitcoin Battles US Bond Nerves With BTC Price Dip Toward New May Lows
ICE and CME urge US regulators to curb Hyperliquid energy trading
On the Scene at Gucci’s Cruise Show in New York City: Mariah Carey, Kim Kardashian, Lindsay Lohan, Iman, and More!
Wall Street’s Boldest Gold Prediction Has Russians Rushing to Buy
IREN closes $3 billion convertible notes deal amid AI infrastructure expansion
Trending Western Style Vests Perfect for Summer
Watch: far-right flag-fanatics run over victim, attack locals – Setup By the Left wing for your entertainment
Holiday Weekend Open Thread – Corporette.com
Blockchain.com files with SEC for U.S. IPO
Amazon Sundays: Memorial Day Hosting
Off Campus Easter Eggs Explained: Characters, Stories, More
Crypto Market Structure Bill Clears Committee; Senate Vote in Focus
BeInCrypto 100 Institutional Awards Nomination: KuCoin for Leader in Digital Asset Adoption and Best Trading Infrastructure
Enphase Energy (ENPH) Stock Rockets 32% This Week: What’s Fueling the Rally?
Voter disenfranchising, bulldozer operations, trampling democracy in India
Strategies for Australian Businesses Entering Southeast Asian Markets
You must be logged in to post a comment Login