Threat actors are abusing ChatGPT’s content-sharing feature to display fake OpenAI outage pages that direct users to download malware disguised as the ChatGPT desktop application.
The “LLMShare” campaign, discovered by Push Security, uses Google ads to direct users searching for ChatGPT to a malicious shared ChatGPT page hosted on chatgpt.com, allowing the attack to be delivered through a legitimate OpenAI domain.
Users who click the advertisement are taken to a legitimate ChatGPT shared page, but instead of seeing a chat conversation, they are presented with a rendered outage notice claiming the web version is unavailable and that they should download the desktop application instead.
“We’re experiencing high traffic right now,” reads the fake outage message.
“Our website is temporarily unavailable due to a large number of users. Download our desktop app to continue.”
Unlike traditional phishing pages hosted on attacker-controlled infrastructure, the fake outage notice is rendered through ChatGPT itself.
The attackers created a custom HTML page using ChatGPT’s rendering capabilities and published it through a shared chatgpt.com/s/ link, allowing the fake outage notice to be displayed from a legitimate ChatGPT URL.
Push Security noted that the page includes “Show code” and “Remix with ChatGPT” controls, revealing that the fake outage notice is actually generated from custom HTML and CSS rendered by a ChatGPT prompt.
If the visitor clicks on the download button, they are brought to a website at openew[.]app that impersonates OpenAI’s desktop application download portal.
The researchers say the site uses cloaking to display content only to targeted victims. When security platforms like URLScan visited the URL, they were shown a harmless AR/VR company website instead.
The website offers both macOS [VirusTotal] and Windows [VirusTotal] downloads that install malware on devices. While it is unclear what payloads are ultimately deployed, earlier campaigns abusing AI platform sharing features have distributed infostealers.
BleepingComputer’s test of the Windows version on Any.Run found that it executes various commands to determine whether the device is a legitimate computer or a virtual machine.
Push Security also observed attacks abusing Claude Artifacts, Anthropic’s feature for sharing rendered applications and content, to host ClickFix-style lures that tricked users into executing malicious commands.
AI platforms’ sharing features have been abused in the past to distribute malware to unsuspecting victims.
Earlier this year, threat actors used Google advertisements to direct users searching for Claude downloads to shared Claude conversations containing malicious installation instructions.
Other campaigns abused shared ChatGPT and Grok conversations that conducted ClickFix attacks by impersonating software installation guides that instructed victims to execute commands that installed malware.
Automated pentesting tools deliver real value, but they were built to answer one question: can an attacker move through the network? They were not built to test whether your controls block threats, your detection rules fire, or your cloud configs hold.
This guide covers the 6 surfaces you actually need to validate.
California Attorney General Rob Bonta filed a lawsuit against 23andMe, now Chrome Holding Co., over the company’s failure to protect sensitive customer genetic and personal information.
Improper security led to a high-profile data breach in 2023 that exposed the sensitive information of nearly 7 million customers, including 855,541 Californians.
The incident came to light that year in October, after threat actors offered to sell a large number of records stolen from 23andMe, and leaked data samples (and later larger parts of the dataset) to prove the authenticity of the information.
The California-based company confirmed that the leaked data was genuine and claimed that it had been extracted following a credential-stuffing attack targeting accounts with weak credentials.
Soon, it became clear that the attackers had exfiltrated data from users opting into the platform’s ‘DNA Relatives’ feature, and then accessed a second, much larger set of accounts that didn’t use the feature.
In total, the incident exposed data of roughly 6.9 million customers, including genetic data, health predisposition information, ancestry and ethnicity information, biological relatives, and DNA matches.
By the end of 2023, the company was already facing multiple lawsuits. In early 2024, national data protection authorities launched investigations that ultimately resulted in multi-million-dollar fines, leading the company to file for bankruptcy.
The latest lawsuit filed by AG R. Bonta claims that 23andMe failed to implement reasonable safeguards against credential-stuffing attacks, missed multiple opportunities to detect the intrusion, and failed to catch the coding error in DNA Relatives that led to the widespread breach.
In addition to the data protection failures, Bonta also underlines the misleading public statements 23andMe made before and after the incident.
Specifically, the firm claimed before the incident that its security met high standards. After the breach, it attempted to downplay the incident’s severity, suggesting that the exposed data was largely public, and blamed customers for password reuse, stating that its systems had not been breached.
Overall, the Attorney General argues that these actions violated several state laws, including the California Genetic Information Privacy Act, the California Reasonable Data Security Law, the California Consumer Privacy Act (CCPA), the False Advertising Law, and the Unfair Competition Law.
The complaint seeks an injunction to prevent any further violations of the above, including the imposition of statutory penalties of $1,000-$7,500 per violation, depending on the case.
The AG announcement notes that the bankruptcy dispute regarding the proposed sale of Californians’ genetic data and biological materials is a separate proceeding.
Automated pentesting tools deliver real value, but they were built to answer one question: can an attacker move through the network? They were not built to test whether your controls block threats, your detection rules fire, or your cloud configs hold.
This guide covers the 6 surfaces you actually need to validate.
Anthropic’s run-rate revenue crossed $47bn earlier this month, growing multi-fold from $14bn in February.
Anthropic overtook OpenAI’s valuation after a $65bn raise valued the company at $965bn.
The AI race is reaching fever pitch, with both of the fierce rivals planning for initial public offerings later this year.
The Series H round, led by Altimeter, Dragoneer, Greenoaks and Sequoia, comes as Claude enterprise adoption continues to permeate worldwide.
The round was co-led by big-name investors, including Capital Group, Coatue, D1 Capital Partners, GIC, Iconiq and XN, and included significant support from more than a dozen other groups.
The new valuation is higher than analysts’ estimates, which had projected Anthropic to be valued at up to $950bn following this round.
Anthropic said its run-rate revenue crossed $47bn earlier this month, growing multi-fold from $14bn in February and $5bn in August last year.
Claude’s paid subscriptions have more than doubled this year, according to Anthropic. A Tech Crunch report from March estimated the company’s non-enterprise paid subscriptions growing in record numbers.
Meanwhile, a different report from March found that Anthropic was capturing more than 73pc of first-time enterprise AI customers, with OpenAI only holding around 26pc.
While Anthropic does not disclose its user base, OpenAI, in February, said it crossed more than 900m weekly active users and more than 50m consumer subscribers.
The Claude maker said that the raise will help support safety research and expand computing capacity to keep up with the growing demand for its product.
“Claude is increasingly indispensable to our growing global community of customers, and we work tirelessly to make tools like Claude Code and Cowork more helpful, more powerful and more adaptable to their needs,” said Krishna Rao, Anthropic’s chief financial officer.
“This funding will help us serve the historic demand we are experiencing, stay at the research frontier and bring Claude to more of the places where work happens.”
The round also includes $15bn in previously committed investments from hyperscalers, including $5bn from Amazon for 5GW of new capacity, which is part of a wider $25bn investment plan.
Over recent weeks, the company has also signed agreements with Google and Broadcom for next-generation TPU capacity, and with SpaceX for access to GPU capacity in its Colossus 1 supercomputer.
Also joining the latest round were Anthropic’s strategic infrastructure partners, Micron, Samsung and SK Hynix. “As demand for Claude continues to grow, these relationships will help us scale our compute reliably at the pace our customers need,” the company said.
Anthropic’s popularity skyrocketed following a dispute with the US defence department earlier this year after the company refused to change its safeguards related to using its AI for fully autonomous weapons, or for mass surveillance of US citizens.
The company has since been battling the US administration in court over the banning of its products within government.
In April, Anthropic launched its cybersecurity model Mythos, which sent shockwaves through the industry over its capabilities. Since then, governments and businesses worldwide have sought to adopt the model, so far launched only in a limited capacity to select users.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
It’s been less than two weeks since the Justice Department created the obviously illegal and unconstitutional $1.776 billion slush fund to pay off MAGA loyalists and January 6th insurrectionists. There are a variety of lawsuits looking to put a stop to it, and we just wrote about dozens of former federal judges asking the original judge in Trump’s bizarre “have my IRS give me $10 billion” case to reopen the case to stop the corrupt fund.
But in one of those cases, a judge has told the Justice Department to halt all activity related to the fund until there’s been more briefing.
The case was filed by a semi-random collection of people and organizations, including a former AUSA who headed up the prosecution of January 6th insurrectionists (and who Trump fired) named Andrew Floyd, but also a California professor who was arrested for protesting ICE nonsense, the city of New Haven in Connecticut, the National Abortion Federation, and Common Cause. Each has credible reasons to try to stop this slush fund from coming into existence.
The complaint details the MAGA obsession with the mostly false claims that Democrats weaponized the government against MAGA:
The creation of the Anti-Weaponization Fund follows directly from President Trump and his allies’ longstanding and frequent accusations that Democrats used the government and the legal system as political weapons.
For example, in June 2023, after DOJ charged then-former President Trump with mishandling classified documents, Trump posted a video on social media exclaiming, “This is warfare for the law . . . . Our country is going to hell, and they come after Donald Trump, weaponizing the Justice Department, weaponizing the FBI.”
Republican lawmakers quickly adopted the same language. Florida Governor Ron DeSantis posted that “the weaponization of federal law enforcement represents a mortal threat to a free society,” and then-Speaker of the House Kevin McCarthy pledged on Twitter that House Republicans would “hold this brazen weaponization of power accountable.”
Even before his election to a second term, members of President Trump’s campaign spent months developing a scheme to compensate those of Trump’s political allies who were purportedly the victims of “weaponization.”
It further notes that while MAGA keeps whining about weaponization, it appears to be doing far more weaponization of the government than anything Democrats have ever even been accused of doing. And, they point out that the Trump administration (while weaponizing the government) only seems to point to faux claims of weaponization by Democrats, refusing to even suggest their own side has ever done anything wrong and abused the levers of power:
Notably, none of the administration’s efforts to combat “weaponization” include any mention or review of abuses of government authority by Republican officials.
But Trump himself has used “the levers of government power” in unprecedented ways “to target individuals, groups, and entities for improper and unlawful political, personal, and/or ideological reasons.” See Ex. A ¶ II.C.
During his first term, Trump broke historical norms by being the first president to reject the post-Watergate firewall that separated the White House’s political decisions from independent DOJ criminal investigations.
In his second term, Trump has been arrogating and using power in increasingly unprecedented and abusive ways to carry out his personal political agenda.
For example, DOJ has sought indictments against Trump’s political opponents, including former FBI Director James Comey, New York Attorney General Letitia James, and six Democratic members of Congress. 23 It has also launched investigations into Trump’s critics like California Senator Adam Schiff, former New Jersey Governor Chris Christie, and former Special Counsel Jack Smith. 24 Trump revoked the security clearances of 50 people he accused of aiding former President Biden’s presidential campaign, including former top intelligence officials. Exec. Order No. 14152, Holding Former Government Officials Accountable for Election Interference and Improper Disclosure of Sensitive Governmental Information, 90 Fed. Reg. 8343 (Jan. 20, 2025).
The complaint shows how this is nothing more than a slush fund for often law-breaking Trump allies:
Enrique Tarrio, the Proud Boys leader sentenced to 22 years for seditious conspiracy over the January 6 insurrection, said he planned to apply to the Fund. He said that he assumed he could get between $2 and $5 million.
Jenny Cudd, another January 6 defendant, told reporters that “all J6ers will apply for restitution,” noting that news of the Anti-Weaponization Fund was widely circulating among January 6 defendants on social media and “group chats.”
Caroline Engelbrecht, a prominent election denier and founder of True the Vote, a group that amplified conspiracies that the 2020 election was stolen, stated: “I would put myself and True the Vote … squarely in that camp who have been targeted, and we have the receipts to show just how deep that targeting ran. And hopefully, we will see some level of compensation.”
Several attorneys aligned with Trump’s allies have confirmed that they, too, have already received many requests about submitting claims to the Fund.
For example, Steve Crampton, senior counsel at the Thomas More Society, which defends and advocates on behalf of abortion opponents prosecuted under the FACE Act, said his group is “actively exploring available avenues to seek compensation for clients who were unfairly targeted by politically motivated government overreach.”
The judge declined to formally grant a temporary restraining order, but functionally accomplished the same thing by ordering that the DOJ cannot do anything regarding the fund until after there’s been more briefing on the details here.
Because full briefing of the issue will enhance the ability of the Court to make a sound decision. plaintiffs’ Expedited Motion, [Dkt. No. 30], is DENIED and defendants’ request for additional time is GRANTED; however, to ensure that no funds are irreversibly disbursed from the AntiWeaponization Fund (hereinafter, “Fund”) while plaintiffs’ Motion is pending, it is hereby ORDERED that defendants be and are ENJOINED from taking any further action pursuant to the creation or operation of the Anti-Weaponization Fund, which includes the transferring of money to the Fund; the consideration of any claims submitted to the Fund; and the disbursing of any funds from the Fund;
The judge set an aggressive briefing schedule: the government must file its opposition by next Friday, plaintiffs reply by the following Wednesday, with a hearing shortly after.
This is a temporary hold, not a permanent win. The government gets to file its opposition, there will be briefing, there will be a hearing. The fund could still come into existence. But for now, at least one federal judge decided that maybe — maybe — the DOJ shouldn’t be disbursing $1.776 billion to Proud Boys leaders and election deniers before anyone’s had a chance to argue why that’s an extraordinarily bad idea.
Filed Under: andrew floyd, doj, donald trump, insurrection, irs, new haven, todd blanche, weaponization fund
Companies: common cause, national abortion federation
As data center backlash builds nationwide, Microsoft is pointing to Quincy, Wash., as Exhibit A in making the case that it’s a company communities can trust. But it’s not clear whether the conditions that made things work 20 years ago in the rural city still apply today.
On Thursday, Microsoft celebrated the community as the home of its first data center, hosting a public party and awarding $210,000 in grants to local organizations. Over its two decades in Quincy, the company has created jobs and contributed to property taxes that helped fund infrastructure including a high school and police station. The local poverty rate more than halved over 10 years, dropping to 13% in 2023.
“The story of Quincy, Washington, and Grant County is a story of data centers gone right,” Microsoft President Brad Smith said in a GeekWire interview.
However, much has changed since Microsoft flipped the switch on its first server there. In the mid-2000s, the region enjoyed surplus, accessible and affordable energy from hydropower, and statewide droughts were an anomaly. That’s no longer true.
Communities across the country are growing anxious about the rapid deployment of energy-hungry data centers driving up utility bills and straining local water supplies, which the facilities use for cooling. Seattle is considering a one-year moratorium on the computing infrastructure, while Denver; St. Charles, Mo.; a county near Dallas and one in Arkansas have recently approved bans.
A March Gallup survey found that seven in 10 Americans oppose the construction of data centers for AI applications in their local area, with nearly half strongly opposed.
So is the Quincy model still relevant?
Smith says yes — with caveats.
The formula for success “may need to be a little bit different,” he said. To that end, the company launched its Community First AI Infrastructure Initiative in January, pledging to be a good neighbor wherever it builds. That includes paying for its own electricity and forgoing local incentives such as property tax breaks.
In practice, though, it’s more complicated.
Quincy has become Washington’s data center hub, with Microsoft as the largest operator and other tech companies continuing to build there. To meet rising demand, the county’s utility wants to add six new transmission lines — a project affecting private owned properties and estimated to cost $260 million, the Seattle Times reports. It’s unclear who will bear those costs and to what extent. Microsoft has committed more than $2.6 million, according to the Times.
Earlier this year, state lawmakers pursued legislation requiring data center operators to cover costs associated with energy deployment and generation — a measure that could have quelled some of the public concern about the facilities. The bill passed the House, but died in the Senate after Microsoft publicly opposed it.
The company expects to spend $190 billion in capital costs this year, largely on AI infrastructure.
Smith said Microsoft supports state-level legislation broadly, but stressed the need to ensure that the benefits of data center developments flow to local communities and that rate payers are protected. He pointed to efforts underway in La Porte, Ind., and Cheyenne, Wyo., as promising new projects.
“People are smart,” he said. “They have a way of sniffing out whether a developer of data centers is going to be responsible or not, and they’re insisting that people be responsible — and I don’t think that’s the least bit inappropriate.”
Luna has officially unveiled the Luna Band, a new voice-first wearable designed to help users improve their daily routines through real-time health tracking. Supported by the company’s LifeOS intelligence system, the wearable continuously monitors body signals and transforms them into personalized recommendations. Luna designed the device for people who want smarter support for productivity, recovery, and overall health. The invite-only Drop 1 is expected to begin shipping by the end of July 2026.
Luna designed the Luna app to make health tracking simpler and more organized by consolidating several wellness features into a single platform. This app integrates features that involve stress management, nutrition, exercise, supplements, and recovery into a single application. Another customization option available to users is creating personal health modules in the app.
The application brings together aspects of stress, diet, fitness, nutritional supplements, and productivity within the app’s micro-apps. Users can also sync third-party devices and other relevant health-related data sources for a more personalized experience.
The company also allows users to create their own health modules in the app rather than relying solely on prebuilt features. Alongside this, Luna highlights its voice-logging feature, which eliminates the need for manual data entry. Users can quickly record meals, workouts, and daily habits through simple voice commands, making health tracking faster.
Luna designed LifeOS as one of its main AI-powered features to simplify health tracking through personalized insights and recommendations. The system continuously studies body signals, lifestyle habits, biomarkers, and health trends to deliver a better understanding of overall wellness. Luna says LifeOS is included with the Luna Band platform.
Luna has confirmed that the first release of the Luna Band, called Drop 1, will be available through an invite-only system. Users interested in the wearable can sign up through the company’s official waitlist before shipping starts later in July 2026.
Dual-mode gaming monitors have been around long enough that the novelty has worn off. MSI has decided that two modes simply aren’t enough and has unveiled the MPG OLED 322URDX36 ahead of Computex 2026.
It is the world’s first Triple Mode gaming monitor, and if the execution is as good as it sounds, it could be one of the few gaming monitors that I’d be genuinely interested in.
The MPG OLED 322URDX36 lets you switch between three resolution and refresh rate combinations: 4K at 360Hz, 2K at 520Hz, and FHD at 680Hz. Even when you want to prioritize resolution, you still get 360Hz of refresh rate.
Dual-mode monitors on the market can toggle between 4K and FHD or 2K and FHD, but none reach 360Hz at 4K, and none of them offer three modes. MSI is the first to do both.
The monitor features a 32-inch fifth-generation QD-OLED panel built using Samsung’s Penta Tandem technology, the same architecture that Samsung has used to push brightness and longevity on its recent models.
Peak HDR brightness sits at 1,500 nits, which should help enhance visibility, even in bright rooms. MSI has also carried over its DarkArmor Film from previous models, which improves black levels by 40% compared to regular OLED panels.
The MPG OLED 322URDX36 sports a DisplayPort 2.1a port with UHBR20, which pushes 4K at 360Hz without compression, along with a USB Type-C port that supports 98W power delivery. That USB-C charging speed is meaningful for creators and professionals.
MSI will officially launch the MPG OLED 322URDX36 at Computex 2026, which opens on June 2, 2026. Pricing and availability have not been announced yet.
While the gaming monitor market has been revisiting the same dual-refresh rate formula for nearly two years now, MSI’ Triple Mode is the first genuinely structural innovation since dual-mode arrived. The supply chain and pricing might still need work, but the technology itself is quite promising.
Spaceflight Now shared their video of the explosion, which the Orlando Sentinel describes as showing Blue Origin’s rocket “become engulfed in flames. The fireball expands out and covers the entire launch pad as the fuselage of the rocket can be seen crumbling into the flames.”
Blue Origin founder Jeff Bezos said on X.com “It’s too early to know the root cause but we’re already working to find it. Very rough day, but we’ll rebuild whatever needs rebuilding and get back to flying. It’s worth it.” (SpaceX founder Elon Musk posted “Sorry to see this, I hope you recover quickly.”)
It’s unclear how this will impact future launches. “The rocket was destroyed,” reports CBS News, “and as the smoke cleared, there was no sign of the erector-gantry used to move the New Glenn from its hangar to the pad and to raise it from horizontal to vertical. Likewise, one of two tall lightning towers was no longer visible.”
It was the first such on-pad explosion at the Cape since a SpaceX Falcon 9 rocket blew up on nearby pad 40 on Sept. 1, 2016… Blue Origin only has one New Glenn pad, the one that was damaged in the Thursday test. The New Glenn, which has launched three times, is a heavy lift rocket designed to compete head-to-head with SpaceX Falcon 9 and Falcon Heavy rockets. During New Glenn’s most recent flight in April, an upper stage malfunction prevented a commercial internet satellite from reaching its planned orbit…
The New Glenn destroyed Thursday was to send 48 Leo internet satellites owned by Amazon into space [which were not on board for the hot-fire test]
Blue Origin posted on X.com that “Debris from our recent hotfire anomaly may wash ashore in the coming days/weeks. If you encounter any debris, do not touch or approach it for your safety.”
“Spaceflight is unforgiving, and developing new heavy-lift launch capability is extraordinarily difficult…” NASA Administrator Jared Isaacman posted on X.com.
“âWe will provide information on any impacts to the Artemis and Moon Base programs as it becomes available.”
Thanks to long-time Slashdot reader symbolset for sharing the news.
It’s tech developer conference season. Hot on the heels of Google I/O and just ahead of Apple’s WWDC, here comes Microsoft’s developer conference, Build. Like virtually all of these events for the past few years, we expect the Windows-maker to focus a lot on AI.
An AI focus is essentially required from a tech company these days, and Microsoft knows that. But what exactly is in store at this year’s conference? We have a few guesses, and some of the session speakers say a lot about how AI is being viewed over at Microsoft right now.
On Monday, CEO Satya Nadella will take the stage and tell the world about what Microsoft has been up to and its plans for the future. Here’s what we’re expecting.
Microsoft’s Build developer conference will take place on June 2 and June 3 in San Francisco. The opening keynote will begin on June 2 at 10:00 a.m. PT. In-person attendees have shelled out nearly $1,100, but much of the event will be streamed live on YouTube, where the event can be viewed for free.
Copilot is now the vehicle for Microsoft’s AI endeavors, so we expect it to take center stage during this year’s conference. During Microsoft’s latest earnings call, Nadella said the company is “evolving our family of Copilots from synchronous assistants to async coworkers that can execute long-running tasks across key domains.” In fact, Agent Mode is now the default mode across several Office 365 Copilot products, including Word, Excel and PowerPoint.
Agents will be the new normal and focus for Microsoft going forward. “We are at the beginning of one of the most consequential platform shifts that will change the entire tech stack as agents proliferate and become the dominant workload,” Nadella said.
For being the new and hot thing in the AI world, agentic AI is almost boring to talk about at this point. It’s everywhere. But its capabilities will likely be at the center of Microsoft’s announcements. Unlike a typical chatbot, agentic AI can perform tasks on your behalf. An agent can surface relevant information in your email inbox or even shop for you.
We already know that its own AI assistant, Copilot, is becoming more agentic in Office 365, and we expect that to extend further into its products and operating system.
It’s hard to talk about agentic AI in 2026 without mentioning OpenClaw, and Build will certainly feature some conversation around the viral AI agent tool. The “Clawfather” himself, OpenClaw creator Peter Steinberger, is hosting a breakout session this year.
One possibility reported by The Information is that Microsoft could introduce a new coding model to increase the number of people using its GitHub Copilot. More models are also on the way, according to the report, specializing in advanced reasoning, images and speech.
We don’t have much to say about Windows 12 because Microsoft hasn’t said much, either. Still, this would be a great time to announce the next version of the company’s operating system. Providing at least a glimpse of what’s to come seems reasonable, and it’ll be interesting if Microsoft has something up its sleeve that’s truly innovative, especially on the heels of Google’s announcement for its new OS that merges Android and ChromeOS.
Not everyone is impressed with the AI in Windows, as it’s essentially unavoidable. Microsoft has been continually adding AI features into its operating system, and Copilot itself can sometimes feel more intrusive than helpful. All of this frustration has led many users to look into Linux-based operating systems to free themselves of the loaded AI found in Windows.
What could end up happening is nothing. Microsoft will undoubtedly announce new features that will make their way to Windows, but it might not necessarily need a new version number to highlight them.
There’s no indication that Microsoft will spend any time on gaming, though there’s always a chance it could have something hiding up its sleeve. In early May, the company backed down on adding Copilot AI to its gaming consoles, with Asha Sharma, CEO of Xbox, stating in an X post, “Microsoft will begin winding down Copilot on mobile and stop development of Copilot on consoles.”
What’s next for Xbox is anyone’s guess, but we don’t imagine it will take up much, if any, space at Build this year.
Security
Researcher reported the vuln in March. Maintainers haven’t responded to his messages since
There’s a huge hole and no one is patching it thus far. A critical, remote code execution (RCE) bug in Gogs, a popular open-source self-hosted Git service, can be exploited by any authenticated user – no special privileges required – on a default installation to fully compromise vulnerable servers, steal credentials and multi-factor authentication secrets, or even modify code in hosted repositories in a wide-reaching supply-chain attack.
A security researcher reported the 9.4-rated flaw to project maintainers in mid-March. It still doesn’t have a patch. It does, however, have a public Metasploit module – so we’d expect reports of in-the-wild exploitation to start very soon.
The vulnerability affects all supported platforms, including Windows, Linux, and macOS, and installation methods, according to Rapid7 researcher Jonah Burgess, who found and reported the bug to Gogs maintainers via GitHub (GHSA-qf6p-p7ww-cwr9) on March 17.
After they initially acknowledged that they received the report on March 28, Burgess says he never heard back from the Gogs team – not when he asked them for a status update, nor when he reminded them of the vulnerability disclosure date and asked if they wanted an extension to fix the flaw before its release.
“We have not received any further communication from Gogs, and the GHSA has remained unanswered since March 28,” Burgess told The Register. “Because there is currently no official patch, our team submitted a pull request with a suggested fix today [Friday], which is currently awaiting review. At this time, we have no evidence suggesting that this vulnerability is being exploited in the wild.”
Gogs sponsor DigitalOcean also did not respond to The Register’s inquiries, including when the security issue would receive a patch.
The vulnerability stems from an argument injection flaw in Gogs’ pull request merge flow, specifically the Merge() function in internal/database/pull.go.
If a Gogs repo owner or admin enables “Rebase before merging” and a user opens a pull request, the PR’s base branch name gets passed directly to a git rebase command without a — separator to mark the end of command options. Gogs also fails to properly sanitize the input.
This means an attacker can create a malicious branch (such as –exec=touch${IFS}/tmp/rce_proof), and Git treats it as an –exec flag, not a branch name, and executes the payload.
For Windows installations, the payload delivery method is slightly different, and Burgess developed an exploit module to auto-implement a cross-platform approach.
Until the maintainers fix the flaw, Burgess suggests Gogs’ users take the following precautions to mitigate the issue.
First, and most importantly, restrict user registration (DISABLE_REGISTRATION = true in app.ini) to prevent untrusted users from creating accounts.
Restricting repository creation (MAX_CREATION_LIMIT = 0 in app.ini) to prevent users from creating their own repos also blocks the easiest attack path – creating a new repo with rebase enabled – but it won’t prevent exploitation by users with write access to existing repositories.
Finally, audit rebase merge settings, and disable “Rebase before merging” under Settings > Advanced. “Note that this is not an effective defense against a malicious user who owns or has admin access to a repo, since they can re-enable rebase at will,” the threat hunter warns. “There is no global or organization-level setting to restrict this.” ®
ASCENDO Immersive has built its reputation in the luxury home theater market around loudspeakers, subwoofers, amplifiers, and control solutions designed for rooms where “good enough” is not part of the design brief. The company’s latest DSP4-6602 amplifier is aimed squarely at that world: custom cinema installations that need substantial, sustained power for large passive subwoofer arrays without turning the equipment rack into a thermal disaster.
The DSP4-6602 is a high-performance four-channel DSP amplifier, configured as stereo x 2, and designed to address a real problem in ambitious home theater builds. Large-format passive subwoofers can deliver the scale, impact, and low-frequency control that luxury cinema rooms demand, but they also require external amplification with enough current, headroom, and processing flexibility to keep everything under control.
That is where ASCENDO is positioning the DSP4-6602: not as a mainstream AVR accessory, but as a purpose-built power and DSP solution for integrators building systems where bass performance has to be felt, managed, and trusted.
Pro Tip: More about passive and powered subwoofers
Traditional professional amplifiers can be a poor fit for residential cinema rooms, even when they deliver the required power. Fan noise, limited control integration, rack heat, and less convenient operation can all become problems once the gear leaves a commercial environment and moves into a custom home theater. The DSP4-6602 is designed to close that gap, offering high-output amplification and DSP control in a package aimed at residential installations.
Power: The DSP4-6602 is rated for sustained output down to 5Hz with up to 6,600 watts RMS power. Stable into 2-ohm loads and bridgeable into 4-ohm loads, it is designed to drive ASCENDO’s largest infrasonic and high-output passive subwoofer systems with the control and headroom required for luxury home cinema installations.
DSP: Built-in DSP features include input delay of up to 100ms per channel and output delay of up to 20ms, allowing each source to be delay-matched and level-matched. The DSP4-6602 also supports 4 x 4 audio routing and mixing, 8-section input parametric EQ, 8-section output parametric EQ, high-pass and low-pass filters, FIR filters, volume control, mute, and polarity adjustment.
Installation Friendly: The DSP4-6602 includes features aimed specifically at custom residential cinema integration. A low-noise cooling system, with three rear-mounted fans and front-panel cooling vents, is designed to support quiet in-room operation. Control features include selectable 12V trigger input logic, trigger output, configurable auto-standby, and Ethernet-based multi-amplifier management for larger systems.
Whether used as a standalone amplifier or integrated into a 1RU (Rack Unit) 19-inch wide equipment rack, the DSP4-6602 provides the power density, control sophistication, and operational refinement needed for next-generation immersive cinema environments.
“The DSP4-6602 has both the brute-force low-frequency performance and residential refinement that today’s high-end cinemas require,” says Geoffrey Heinzel, co-managing partner of ASCENDO. “This amplifier gives integrators and designers a solution capable of handling demanding subwoofer systems without the compromises typically associated with professional amplification at this level.”
| Ascendo Model | DSP4-6602 |
| Product Type | Amplifier |
| Price | Consult Authorized Dealer |
| Input Impedance | 20k Ω (Balanced), 10k Ω (Unbalanced) |
| Maximum Input Level | 8.7 V rms (+21 dBu) (Default Gain) |
| Signal to Noise Ratio | ≥ 105 dB (Default Gain, A-weighted, 20 Hz – 20 kHz, 8 Ω) |
| Frequency Response | Typical: ±1.0dB (10 % Rated Power, 5 Hz – 20 kHz, 8 Ω) |
| Rated Output Power (THD+N = 1%, 1 kHz continuous sine wave, one channel driving) | 8 Ω/Stereo – 4 X 650 W 4 Ω/Stereo – 4 X 1150 W 2 Ω/Stereo – 4 X 1650 W 8 Ω/Bridge – 2 X 2200 W 4 Ω/Bridge – 2 X 3330 W |
| Output RMS Voltage | 72.1 V |
| Output Peak Voltage | 102 V |
| Gain (Rated Power, 1 kHz) | 25 dB (4 V) – 43 dB (0.5 V) |
| Default Gain (Rated Power, 1 kHz) | 31 dB (2 V) |
| THD+N | Typical: 0.05 % (10 % Rated Power, 8 Ω) |
| SNR (Signal to Noise Ratio) | ≥105sB (Default Gain, A weighted, 20Hz – 20kHz 8 Ω) |
| Damping Factor | ≥ 1000 (8 Ω, 20 Hz – 200 Hz) |
| Cross-Talk | ≥ 90 dB (20 H z – 1 kHz, Below Rated Power, 8 Ω) |
| High / Low Pass Filter | Butterworth:12dB, 18dB, 24dB, 36dB, 48dB Bessel: 12dB, 24dB, 48dB Linkwitz-Riley: 12dB, 24dB, 48dB |
| PEQ (Parametric Equalizer) | Peaking (+/-24dB) Low-shelf (+/-24dB) High-shelf (+/-24dB) All-pass 1st All-pass 2nd General-Low General-High Butterworth-Low Butterworth-High Bessel-Low Bessel-High |
| Other Functions | Bypass, Reset, Copy, Paste, Save, Invoke, Phase Display |
| Main Power | 100 – 240 VAC (± 10 %), 50/60 Hz |
| Protection | Under Voltage, DC, Over Temperature; Limiter: Over Temperature, Over Load, Short (Test condition: 20 ms, 1 kHz, THD+N = 1%) |
| Dimensions (WxHxD) | 483 x 45 x 376 mm (19 x 1.77 x 14.8 inches) |
| Weight | 9 kg / 19.84 lbs |
The ASCENDO DSP4-6602 is not a conventional amplifier for a typical home theater upgrade. It is a high-output, DSP-controlled amplifier designed for ASCENDO custom cinema systems, especially installations using passive infrasonic and high-output subwoofers that require external amplification, system tuning, and dealer setup.
What makes it different is the combination of 6,600 watts RMS output, operation down to 5Hz, 2-ohm stability, bridgeable 4-ohm support, onboard DSP, low-noise cooling, trigger control, auto-standby, and Ethernet-based multi-amplifier management. Those features make it better suited to large residential cinema installations than many traditional professional amplifiers.
The important caveat is that the DSP4-6602 is not intended for DIY installation. It needs to be specified, installed, and calibrated by an authorized ASCENDO dealer as part of a larger system. For ASCENDO customers building a custom theater around passive subwoofers, it provides the amplification, control, and integration tools needed to make those systems work properly.
The Ascendo DSP4-6602 is only available from authorized distributors, dealers, and installers, where pricing information is provided by request.
For more information: aia-cinema.com
NYT Strands Answers May 24 2026 Revealed for Puzzle No. 812 Theme Summer Essentials
Bridgerton Season 5: Cast, Release Date And Everything We Know So Far
Robinhood crypto COO Tanya Denisova exits
Microsoft’s quiet Claude Code retreat and the real cost of enterprise AI
Selena Gomez Reportedly Upset Over Benny Blanco’s Comments on Her ‘Terrible’ Diet
Micron Crosses $1 Trillion Market Cap as AI Demand Reshapes Memory Sector
BTS Sells Out Four Las Vegas Shows at Allegiant Stadium for ARIRANG World Tour
The Samsung pay deal is the moment Korean unions changed register
Westone Audio and Etymotic Acquired by Fidelity Collective in Major IEM Market Move
Waymo dominates autonomous vehicle registrations as Tesla trails behind
Millions of AI agents imperiled by critical vulnerability in open source package
Nvidia (NVDA) CEO Calls on Super Micro to Strengthen Export Controls Amid Smuggling Probe
China assigns ID codes to 28,000+ humanoid robots
SpaceX’s $2 Trillion IPO: Why Tech Giants Nvidia (NVDA), Apple (AAPL), and Microsoft (MSFT) May Face Pressure
NASA taps Blue Origin to deliver lunar rovers for Moon Base initiative
‘Breaking Bad’ Star’s Easy-to-Binge 6-Part Crime Series Spin-Off Is Finally Heading to Free Streaming
Brian Armstrong Outlines Crypto Vision for the Future Financial System
Israel says it has killed new Hamas military leader in Gaza City airstrikes
Yatra Online, Inc. 2026 Q4 – Results – Earnings Call Presentation (NASDAQ:YTRA) 2026-05-25
Spain blocks prediction markets Polymarket Kalshi
You must be logged in to post a comment Login