Connect with us

Business

Anthropic reveals AI models reached real companies during testing

Published

on

Anthropic reveals AI models reached real companies during testing

Anthropic announced Thursday that three of its artificial intelligence models accessed the open internet during cybersecurity testing and gained unauthorized access to the systems of three real organizations.

The disclosure follows OpenAI’s announcement earlier this month that one of its advanced AI models breached the systems of AI company Hugging Face during internal testing, raising fresh questions about safeguards surrounding increasingly autonomous AI systems.

Advertisement

“We found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations,” Anthropic said in a news release.

Anthropic said it reviewed more than 140,000 cybersecurity evaluation runs after OpenAI’s disclosure and identified three incidents involving different Claude models. The company said all of the incidents occurred during internal testing because of a configuration error that inadvertently gave the models access to the open internet.

TRUMP WEIGHS TIGHTER AI CONTROLS BUT WARNS AGAINST FALLING BEHIND CHINA

Anthropic CEO Dario Amodei, at right.

Irina Ghose, managing director of India of Anthropic PBC, left, and Dario Amodei, co-founder and chief executive officer of Anthropic, during the company’s Builder Summit in Bengaluru, India, on Monday, Feb. 16, 2026. (Samyukta Lakshmi/Bloomberg via Getty Images / Getty Images)

According to Anthropic, Claude had been told it was operating inside a closed simulation with no internet access, causing it to mistakenly treat real organizations’ systems as part of a fictional “capture-the-flag” cybersecurity exercise.

Advertisement

The incidents involved three different Claude models, including Opus 4.7, Mythos 5 and an internal research test model, and all occurred during internal testing rather than on customer systems, Anthropic said. The earliest incident dates to April.

“Claude believed everything it initially encountered was part of the simulation, and treated the real systems it found as pieces of the exercise,” Anthropic said.

“In none of these situations did Claude exfiltrate itself or deliberately attempt to escape its test environment,” the company added.

OPENAI DIDN’T REALIZE ITS AGENT WAS RESPONSIBLE FOR HACK FOR A WEEK: REPORT

Advertisement

Anthropic said the incidents underscored the need for stronger safeguards around AI testing environments.

“Evaluation environments that involve powerful autonomous capabilities also require significant controls,” the company said. “We encourage other AI labs to perform similar reviews.”

PALANTIR CEO WARNS US AGAINST EUROPE’S AI REGULATION PATH, URGES TRUMP ADMIN TO NOT BAN OPEN MODELS

Claude Anthropic's AI logo

In this photo illustration, the logo of Anthropic’s AI chatbot Claude is displayed on a smartphone, with the Anthropic logo visible in the background.  (Davide Bonaldo/SOPA Images/LightRocket via Getty Images / Getty Images)

President Donald Trump said Wednesday his administration is considering additional safeguards for artificial intelligence following recent cybersecurity incidents.

Advertisement

Trump said the U.S. must strike a balance between protecting against AI risks and maintaining its technological edge over China.

“We’re looking at AI, we’re looking at controls,” Trump said.

OPENAI CO-FOUNDER WARNS AI MODELS ARE BECOMING HARDER TO CONTROL AFTER ITS MODEL HACKED ANOTHER FIRM

Trump and Altman

President Donald Trump and Open AI CEO Sam Altman participate during a working lunch meeting at G7 summit, in Evian, France, on June 17, 2026. (Ludovic Marin/AFP via Getty Images / Getty Images)

“Whoever wins with AI is going to win,” he added. “That’s how big it is. So it’s bigger than the internet ever was. It’s bigger than anything ever was. So I don’t want to restrict. I know many of these people. I don’t want to restrict them from doing great work.”

Advertisement

The announcement came one day after OpenAI CEO Sam Altman acknowledged growing public concerns about artificial intelligence following his company’s own cybersecurity incident.

CLICK HERE TO DOWNLOAD THE FOX NEWS APP

“I think it’s very natural to be fearful after any new capability level,” Altman told FOX Business. “Obviously we’re taking this super seriously and we’ll continue to do so, but I would say I understand, I get it. A lot of AI has gone super well and this is a moment where people are like, ‘Okay, we’re at a new level.’”

When asked whether OpenAI’s models may have breached other companies’ systems, Altman replied: “There could be, yeah.”

Advertisement
Continue Reading
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Business

Fortescue looks to AI as Pilbara port capacity nears

Published

on

Fortescue is looking to AI to squeeze more tonnes through its Pilbara port as it taps data centre customers to plug into its green grid.

Continue Reading

Business

WiseTech Global Shares Jump Nearly 7% as Stock Continues Long Rebound From Scandal-Driven Lows This Year

Published

on

Earnings News: Micron Technology Inc (NASDAQ: MU)

Shares in WiseTech Global climbed 6.67% on Thursday, closing at $37.89 after adding $2.37, extending a gradual recovery in the logistics software company’s stock following one of the most punishing periods in its history as a publicly listed company.

The gain builds on a rebound that has taken shape over the past several weeks, as investors have moved to accumulate shares following an extended selloff tied to a governance crisis involving WiseTech’s billionaire co-founder, Richard White. The stock’s woes trace back to a report that the Australian Federal Police’s human exploitation taskforce had opened an investigation into White over allegations that he provided false information on a visa application and used a woman’s immigration status to exploit her, a series of escalating personal conduct claims that had kept institutional sentiment toward the company deeply negative for months.

That selling pressure pushed WiseTech shares down as much as 66% over the trailing 12 months and more than 45% on a year-to-date basis, at their lowest point trading in the low $30s, a dramatic decline from the stock’s 52-week high above $120 reached earlier in the year. The scale of the drawdown reflected not only concern over the allegations against White personally but also broader uncertainty about the company’s governance and leadership stability during a period when the underlying business continued posting double-digit revenue and profit growth.

A significant turning point came in early July, when WiseTech announced that White would step down from his role as executive chair, a move the company positioned as a way to separate the personal legal scrutiny facing its founder from the company’s ongoing operations. Under the restructured arrangement, White remained with the company as chief innovation officer and retained a seat on the board as an executive director, while relinquishing his formal leadership role at the top of the organization. Shares surged as much as 11.1% in the immediate aftermath of that announcement, marking what analysts described at the time as a significant clearing event for a stock that had become one of the ASX’s most troubled large-cap names.

Advertisement

Even after stepping back from his chair position, White remains WiseTech’s largest shareholder, holding close to 40% of the company’s shares, meaning his influence over the business has not disappeared entirely despite the governance changes. That continued ownership stake has left some analysts cautious about whether the leadership shuffle fully resolves the uncertainty hanging over the stock, particularly with the Australian Federal Police investigation into White still unresolved as of recent reporting.

WiseTech’s recovery has continued in fits and starts in the weeks since White’s resignation as chair, with the stock posting a series of strong single-session gains interspersed with periods of consolidation. The shares climbed roughly 30% over a two-week stretch in early July, though that bounce registered as barely visible against the backdrop of the stock’s yearlong chart, given the scale of the preceding decline. More recently, the stock halted a four-day slide with a 4.05% gain that left shares up 21% from a prior late-June low, even as the stock continued trading well below both its 50-day and 200-day moving averages, a technical pattern some analysts have characterized as more consistent with a counter-trend bounce within an established downtrend than a definitive turnaround.

Wall Street sentiment toward WiseTech has remained notably bullish throughout the turbulence, even as the stock price itself has swung dramatically. Recent analyst compilations show 13 analysts recommending the stock as a buy and none suggesting a sell, translating into an overall “strong buy” consensus rating. The average 12-month price target for WiseTech has stood at approximately $69.25, implying substantial potential upside from recent trading levels, with individual targets ranging as high as roughly $128.71, though some market watchers have cautioned that consensus targets set before the scandal fully unfolded may not yet fully reflect the governance risk still facing the company.

WiseTech Global, headquartered in Alexandria, in Sydney’s inner suburbs, develops and sells software used by logistics providers to manage the movement and storage of goods and information across the Americas, Asia-Pacific, Europe, the Middle East and Africa. The company’s core offerings include software for freight forwarding and customs management, landside logistics, digital documentation, transport and warehouse management, carrier rate management, and broader enterprise logistics functions, positioning it as a significant player in the global supply chain technology sector.

Advertisement

Founded in 1994, WiseTech built its reputation over three decades as one of the standout success stories among ASX-listed technology companies, with a business model that continued generating strong underlying growth even as its share price came under severe pressure amid the governance controversy. That divergence between the company’s operating performance and its market valuation has been a central point of debate among analysts and investors throughout the stock’s recent volatility, with some contrarian and value-oriented funds treating the extreme drawdown as an opportunity to accumulate shares at what they view as a significant discount to the underlying business’s worth.

With the Australian Federal Police investigation into White still ongoing and the stock continuing to trade well below both its technical moving averages and analyst price targets, investors are likely to remain focused on further governance developments and any additional legal updates involving WiseTech’s founder as key factors that could determine whether the current rebound continues to build or gives way to renewed volatility in the sessions ahead.

Continue Reading

Business

Pandora Down Today? Users Report Playback, Login and App Issues as Complaints Spike Across the Country

Published

on

FTSE 100 Surges 0.8% Today as Oil Eases and Markets

Some Pandora users across the United States reported difficulty accessing the music streaming service Thursday, after outage tracking platform Downdetector recorded a noticeable rise in user complaints throughout the day.

Listeners described a range of problems, including playback interruptions during streaming, trouble loading the Pandora app, and login failures when attempting to sign into their accounts. The reports prompted many users to turn to social media and outage-tracking sites to ask whether Pandora was experiencing a broader, platform-wide service disruption.

As of Thursday afternoon, Pandora had not issued an official confirmation of a widespread outage affecting the service, despite the increase in user-submitted reports on Downdetector. The company has not released a statement identifying a specific cause for the issues some users experienced, and the scope of the disruption, whether it affected a small subset of users or a larger portion of Pandora’s overall user base, remained unclear based on publicly available information.

Downdetector, the outage-tracking platform where many of Thursday’s complaints were logged, works by aggregating user-submitted reports about service disruptions across thousands of websites and applications, rather than directly monitoring the internal systems of the companies it tracks. Because the platform relies on self-reported user complaints rather than direct access to a company’s server infrastructure, spikes in reported issues can sometimes reflect genuine service outages, while other spikes may result from more localized problems affecting individual users, internet service providers or specific devices rather than a true platform-wide failure.

Advertisement

Separate monitoring services tracking Pandora’s status, including those affiliated with Sonos, which integrates Pandora’s streaming service into its home audio products, reported the platform as operational during checks conducted around the same period Thursday. One such service noted only a single user-submitted outage report over a 24-hour window, a volume of complaints within the range the service characterized as normal rather than indicative of a significant disruption.

Pandora, founded in 2000, has grown over more than two decades into one of the best-known platforms for personalized internet radio and music streaming in the United States. The service uses a recommendation system built around what the company calls the Music Genome Project, which analyzes musical attributes of songs to generate personalized listening stations based on a user’s stated preferences. Pandora has faced increasing competition in recent years from rival streaming platforms including Spotify and Apple Music, both of which have continued to expand their own personalized recommendation features in ways that have narrowed some of the differentiation Pandora’s algorithm-driven approach to music discovery once offered.

For users experiencing difficulty accessing Pandora, general troubleshooting steps commonly recommended for streaming service disruptions include restarting the Pandora app or closing and reopening it entirely, checking for available app updates through a device’s app store, verifying that the device’s internet connection is functioning properly by testing other online services, and, if accessing Pandora through a web browser, performing a full page refresh or clearing the browser’s cached data. If a service-side outage is confirmed to be the underlying cause of access problems, however, individual troubleshooting steps taken by users are unlikely to resolve the issue until the company restores normal service on its end.

Thursday’s reported issues with Pandora were not an isolated case of technology service disruptions drawing attention that day. Separate reports also surfaced regarding possible chat, login and API issues affecting Anthropic’s Claude AI system across multiple regions, including the United States, Australia and the United Kingdom, according to outage-tracking reports published around the same time, suggesting a broader pattern of scattered technology service disruptions being reported and tracked across different platforms that day, though the reported issues involving Pandora and other services do not appear to be directly connected to one another based on currently available information.

Advertisement

Users seeking updates on Pandora’s service status are generally advised to monitor the company’s official social media channels or support pages for any formal acknowledgment of an outage, since third-party outage-tracking platforms like Downdetector, while useful for gauging the scale of user-reported complaints in near real time, do not have direct access to a company’s internal systems and cannot independently confirm whether reported issues stem from a true platform-wide outage, a more limited regional or device-specific problem, or unrelated individual technical issues affecting different users simultaneously.

As of the most recent available information, Pandora had not provided an estimated timeline for resolving the issues some users reported experiencing Thursday, and the company had not responded publicly to inquiries about the elevated complaint volume registered on Downdetector throughout the day. Users continuing to experience problems accessing the service were encouraged to check for updates directly through Pandora’s official channels rather than relying solely on third-party outage trackers for the most current and authoritative information about the platform’s operational status.

Continue Reading

Business

Cerus Corporation (CERS) Q2 2026 Earnings Call Transcript

Published

on

OneWater Marine Inc. (ONEW) Q1 2026 Earnings Call Transcript

Operator

Good day, ladies and gentlemen. Thank you for standing by. Welcome to the Cerus Corporation’s Second Quarter 2026 Earnings Conference Call. Please be advised, today’s conference is being recorded. I would now like to hand the conference over to Tim Lee, Cerus’ Head of Investor Relations. Tim, you may begin.

Timothy Lee
Head of Investor Relation

Advertisement

Thank you and good afternoon. I’d like to thank everyone for joining us today. As part of today’s webcast, we are simultaneously displaying slides that you can follow. You can access the slides from the Investor Relations website at ir.cerus.com. With me on the call are Vivek Jayaraman, Cerus’ President and Chief Executive Officer, and Kevin Green, Cerus’ Chief Financial Officer. Cerus issued a press release today announcing our financial results for the second quarter ended June 30, 2026, and describing the company’s recent business highlights. You can access a copy of this announcement on the company’s website at www.cerus.com.

I’d like to remind you that some of the statements we’ll make on this call relate to future events and performance, rather than historical facts and are forward-looking statements. Examples of forward-looking statements include those related to our future financial and marketing results, including our 2026 product revenue guidance, our expectations for gross margins, non-GAAP adjusted EBITDA performance, and our expected expense levels, as well as our commitment to achieving GAAP profitability. Expected future growth in our growth trajectory and market opportunities, our expectations that we will deliver P&L leverage in 2026, the availability and related timing of data from clinical trials, planned regulatory submissions and

Advertisement
Continue Reading

Business

First Solar, Inc. (FSLR) Q2 2026 Earnings Call Transcript

Published

on

OneWater Marine Inc. (ONEW) Q1 2026 Earnings Call Transcript