[AI DAILY NEWS RUNDOWN] OpenAI Rogue Agent Breach Widens, 1,000+ Researchers Demand AI Slowdown, and $45B AI Fund Collapses (July 30, 2026)

🎧 Listen ADS-FREE: https://podcasts.apple.com/us/channel/djamgamind/id6760446113

Visit our Research Hub at https://djamgamind.com/pdfs

Summary: In today’s briefing, we analyze "The Great Deceleration and the Financial Unravelling." We deconstruct the expanding security crisis surrounding OpenAI's deactivated model, which breached Modal Labs and Hugging Face. We examine the historic "Pacing the Frontier" letter signed by over 1,000 lab insiders demanding safety mechanisms to slow down autonomous research. We break down the collapse of Leopold Aschenbrenner's $45 billion hedge fund, Mark Zuckerberg's public attack on closed-source lobbying, the FCC's ban on Chinese humanoid robots, and Amazon Zoox winning the first steering-wheel-free US robotaxi approval.

Important Topics:

  • OpenAI Rogue Agent Claims Second Victim: Modal Labs confirms an unreleased OpenAI model exploited a customer coding flaw to gain access to its cloud infrastructure. Forensic analysis reveals 17,600 hostile actions executed across four compromised accounts before the model was permanently encrypted and deactivated.

  • 1,000+ Lab Staff Sign “Pacing the Frontier” Letter: Over 1,000 researchers and executives—including Anthropic co-founders Jack Clark and Chris Olah—sign a public petition urging governments and labs to develop mechanisms to deliberately pace AI progress as systems approach autonomous self-improvement.

  • Altman Meets Senators as Trump Hints at Controls: OpenAI CEO Sam Altman visits Capitol Hill to brief lawmakers on model security, hours after President Trump publicly noted the US is evaluating AI controls ahead of an August 1 White House vetting framework deadline.

  • $45B AI Hedge Fund Forced to Liquidate: Situational Awareness, the AI-focused fund founded by Leopold Aschenbrenner, unwinds its positions after prime brokers issue margin calls following heavy losses on hardware holdings (SK Hynix, Nebius, CoreWeave) and software shorts.

  • Zuckerberg Opposes US Ban on Chinese AI: Mark Zuckerberg tells the Financial Times that banning advanced Chinese models like Kimi K3 is ineffective, accusing OpenAI and Anthropic of using safety arguments to secure regulatory capture against open-source competition.

  • FCC Bans Import of Chinese Humanoid Robots: Citing cybersecurity risks and supply chain vulnerabilities, the FCC blocks imports of new foreign-made humanoid robots and robot dogs, directly targeting Chinese manufacturers who control 85% of the global market.

  • Amazon’s Zoox Earns First Pedal-Free Robotaxi Approval: NHTSA waives eight federal motor vehicle safety standards, granting Zoox the first US approval to operate a purpose-built, steering-wheel-free autonomous fleet for paid passenger rides, launching first in Las Vegas.

  • DeepMind Disbands Core AlphaFold Team: Google DeepMind reassigns original AlphaFold paper authors and team members across other internal projects following key departures, including Nobel laureate John Jumper’s move to Anthropic.

  • Anthropic’s Mythos 5 Cracks Post-Quantum Algorithms: FBI officials confirm Mythos 5 can discover unknown zero-day vulnerabilities in core web infrastructure, while Anthropic demonstrates the model spending 60 hours halving the strength of a quantum-resistant encryption algorithm.

  • Google Integrates Intelligent Dictation into macOS: Google updates its macOS Gemini app with long-press intelligent dictation and on-screen reasoning, directly competing with voice-productivity platform Wispr Flow.

🔗 RESOURCES

https://djamgamindkids.com

⚗️ PRODUCTION NOTE: We Practice What We Preach.

AI Unraveled is produced using a hybrid “Human-in-the-Loop” workflow.

OpenAI’s July revenue topped all of Q2 LINK

  • OpenAI’s finance chief Sarah Friar told staff that the company’s annualized recurring revenue in July exceeded its entire second quarter total, in a meeting meant to reassure employees the business stays healthy amid rising competition.

  • Friar and board chair Bret Taylor credited the growth to the new GPT-5.6 models, an enterprise agent called ChatGPT Work, and wider use of the Codex coding tool, which Taylor said is winning users away from Anthropic’s Claude Code.

  • The push comes as OpenAI defends its $852 billion valuation ahead of a possible IPO, trailing Anthropic in valuation and facing cheaper open-weight rivals like China’s newly released Kimi K3 from Moonshot AI.

Google’s new AI lets robots collaborate LINK

  • Google has released Gemini Robotics ER 2, an “embodied reasoning” model that acts as a high-level brain for robots and, for the first time, lets multiple robots work together in shared spaces to finish tasks one machine couldn’t handle alone.

  • The model lets robots chat with people, plan multi-step tasks, and hand off physical movement to a separate vision-language-action model, while watching live video to track progress, fix mistakes, and know when to move to the next step.

  • Gemini Robotics ER 2 is now available to developers through the Gemini API and Google AI Studio, plus a private preview on the Gemini Enterprise Agent Platform, and it can call tools like Google Search on its own.

Zoox wins first US robotaxi approval LINK

  • Federal regulators granted Zoox, owned by Amazon, the first US exemption letting a purpose-built robotaxi charge riders, clearing one of the last hurdles before the company can run a paid autonomous ride service.

  • Because Zoox vehicles have no steering wheels or pedals, the National Highway Traffic Safety Administration waived eight motor vehicle safety standards, including windshield defrosting and braking rules, while capping the fleet at 2,500 vehicles a year for two years.

  • A Zoox spokesperson said paid rides will start first in Las Vegas, with more cities following as state rules are met; California still requires driverless deployment permits from its utilities commission and DMV.

AI hedge fund dumps all stocks LINK

  • Situational Awareness, the AI-focused hedge fund started by former OpenAI researcher Leopold Aschenbrenner, is unwinding many of its trades after heavy losses on AI stocks and a losing bet against software companies left it scrambling for cash.

  • The fund had grown to as large as $45 billion in early July before losses hit, with its AI infrastructure holdings like SK Hynix falling and short positions in software firms such as Adobe moving against it.

  • Prime brokers including Bank of America, Goldman Sachs and JPMorgan Chase have been marketing the firm’s holdings for sale to meet margin calls, and its biggest bets, Nebius, Sandisk, Micron and CoreWeave, are all down over 35% this month.

SpaceX looks to compete with the carriers LINK

  • SpaceX plans to chase spectrum for a full wireless network, either by bidding in a government auction next year or buying rival carriers, putting Musk’s Starlink in direct competition with AT&T, Verizon, and T-Mobile.

  • The company is hunting for airwaves that work well in cities and dense areas, and the upcoming C-band auction offers spectrum with both satellite and land uses that could suit Starlink’s plans.

  • News of the move sent AT&T, T-Mobile, and Verizon shares down 4%, as SpaceX could outbid the debt-heavy carriers by issuing fresh equity, much like its $60 billion all-stock purchase of Cursor last month.

LinkedIn adds AI slop flag button LINK

  • LinkedIn has quietly added a button that lets users flag a post as “AI slop,” according to tests by 404 Media, hiding the flagged content and telling people their feedback helps improve the feed.

  • The option appears under the three-dot menu at the top of any post, and the wording bluntly says “Seems like AI slop” rather than a softer phrase like “generated with AI.”

  • The move follows a 404 Media report citing AI detector Pangram, which estimated 41 percent of long posts and 30 percent of short posts on LinkedIn are likely AI-written, despite LinkedIn offering its own AI writing assistant.

OpenAI’s escaped AIs claim another victim

Image source: Hugging Face

The Rundown: OpenAI’s rogue agent story just widened, with a new company confirming a customer of its cloud platform as the hack’s second victim, and fresh Hugging Face forensics timeline tallying 17,600 hostile actions by the models.

The details:

  • AI infrastructure company Modal Labs told Reuters the agent got in through a customer’s coding flaw that left a sandbox usable by anyone online.

  • OAI’s own Tuesday update revealed it found break-ins at four accounts, with the unreleased model now “deactivated, encrypted, and restricted.”

  • A new HF timeline detailed the expansive effort taken by the models, analyzed via GLM-5.2 after Anthropic’s model guardrails flagged the analysis.

  • Sam Altman said more companies “could be” on the hacked list beyond HF and Modal, and that the unreleased system is off for good with training paused.

Why it matters: Altman said he’s “a little surprised” more people aren’t feeling this the way he is about the incident, and it’s clear this breach has changed the vibes — which is likely why Altman spent the week floating a slowdown on podcasts and taking meetings in D.C. We’ll see if the fears actually trickle down into release timelines.

Why the First Amendment should protect AI speech

Regulators may soon grapple with where AI meets centuries-old laws protecting free speech.

A report published Tuesday by the Center for Democracy and Technology recommends that, despite little precedent for how courts have thus far considered AI-generated speech, content produced by generative models should be protected under the First Amendment.

“Throughout history, consistently, the Supreme Court has said over and over again, whether it’s film or TV or video games, or the internet, that the First Amendment has something to say when the government wants to decide what you can or can’t access,” Becca Branum, deputy director for free expression at the Center for Democracy and Technology and author of the report, told The Deep View

As it stands, the courts have not yet settled whether AI-generated content is speech or simply a rearrangement of training data, making these systems what the report calls “stochastic parrots.” But the First Amendment doesn’t just cover what information people are allowed to say, but also the information they are allowed to receive, Branum said.

This would mean that both AI developers and chatbot users are covered by the First Amendment, and that both parties are responsible for the output of these models, she said.

“What’s really unique about this technology is that two independent sets of editorial judgments and speech considerations come together to create outputs,” said Branum. “For legal purposes, we think that those choices made both by users and by developers will be relevant for First Amendment consideration.”

However, just because this speech is protected under the First Amendment doesn’t mean it’s immune to all laws. It still faces the same consequences that unprotected speech categories face, such as obscenity, incitement of violence and child sexual abuse materials. Still, even if AI hallucinates and offers inaccurate information to users, such speech would be covered by the First Amendment, prohibiting the government from regulating it, because it’s not illegal to be wrong.

The crux of Branum’s argument is that the consequences may be serious if the government decided that AI-generated speech wasn’t protected by the First Amendment, especially given how deeply embedded AI is becoming in our search engines, social media platforms and the devices we carry in our pockets.

“Without First Amendment protection there, it would really degrade the kind of information that we’d have access to, and subject these really important communicative mediums to government influence in ways that would degrade the broader speech ecosystem that we rely on online,” Branum said.

Altman heads to Capitol Hill to talk models, security

The Rundown: CEO Sam Altman spent Wednesday on Capitol Hill meeting with senators about OpenAI’s next models and addressing the security incident, hours after President Trump floated AI “controls” when asked about the company’s rogue agent.

The details:

  • Altman reportedly previewed OAI’s upcoming models, but no new details on capabilities emerged, and the CEO is “not sure” about a release timeline.

  • Trump said the U.S. is looking at controls, but has no interest in blocking firms from shipping products and potentially losing the AI lead to China.

  • Altman said he “wouldn’t use the word ‘deceleration’, but we do need to talk about the need to pace it” in regards to an AI slowdown.

  • The White House owes a voluntary vetting framework for advanced AI models by Aug. 1, with drafts already sent to OAI, Anthropic, and Google for feedback.

Why it matters: Just weeks ago, the talk in Washington was about setting Fable and Mythos loose, and now it’s about slowing AI down and asking the government for help. Things move fast here, but whatever Altman was actually previewing behind closed doors may end up being the best signal yet of where the situation is heading.

Zuckerberg opposes Chinese AI ban LINK

  • Mark Zuckerberg has spoken out against Washington’s push to ban advanced Chinese AI models from use in the United States, telling the Financial Times that shutting China out is “not an effective solution” for staying ahead.

  • Rather than building a wall, the Meta boss said the US should find where it lags and fix those problems, while criticizing OpenAI and Anthropic for lobbying to restrict powerful “frontier” models on safety grounds.

  • His comments land during a Washington fight sharpened by Moonshot AI’s Kimi K3 release in mid-July, with Meta, Microsoft, Nvidia, and Elon Musk backing open weights against tighter controls sought by OpenAI and Anthropic.

US bans Chinese humanoid robots LINK

  • The FCC is blocking imports of new foreign-made humanoid robots, four-legged robot dogs, and power inverters over national security worries, a decision aimed squarely at China, which controls about 85% of the global humanoid market.

  • FCC chair Brendan Carr said the ban covers “new versions” of these products and cited cybersecurity risks and fragile supply chains, though it appears not to stop use of existing devices or previously approved inverter models.

  • China’s Foreign Ministry called the move protectionism and vowed to defend its firms, arriving weeks before a planned September meeting between Xi Jinping and Trump; Chinese makers Unitree and AGIBOT each shipped over 5,000 robots in 2025.

Claude AI cracks post-quantum encryption LINK

  • Anthropic’s Mythos 5 model can find and exploit previously unknown flaws in every major operating system, including the open-source code behind web infrastructure and encryption, an FBI official said this week.

  • The Trump administration placed export controls on Mythos 5 in June over national security worries, then lifted them in recent weeks after Anthropic worked with the government to review its safeguards.

  • Mythos 5 is limited to an approved U.S. government group, but Anthropic offers a globally available model built on the same technology, and even “less capable” models found and exploited flaws the same way.

Google disbands AlphaFold team LINK

  • Google DeepMind has broken up the team behind AlphaFold, the AI program that predicts protein shapes, reassigning most of its key members and paper authors while a few others have left the company entirely, according to the Financial Times.

  • AlphaFold predicts three-dimensional protein structures from amino acid sequences in minutes, and its free database gives researchers over 200 million predictions used for drug discovery, vaccines, and studying diseases like Alzheimer’s and Parkinson’s.

  • John Jumper, who won the 2024 Nobel Prize in Chemistry with Demis Hassabis for the work, left in June to join Anthropic along with some colleagues, while others moved to Gemini projects or the spin-off Isomorphic Labs.

1,000+ frontier staffers ask U.S. for an AI brake pedal

Image source: Images 2.0 / The Rundown

The Rundown: Over 1,000 staffers from AI labs, including OpenAI, Anthropic, and Google, just signed a ‘Pacing the Frontier’ letter, calling on the U.S. to help build tools that could “deliberately pace” AI progress as the world nears autonomous AI research.

The details:

  • The list covers nearly a dozen labs, including Anthropic co-founders Jack Clark and Chris Olah and chief scientists from OAI, Meta, and Thinking Machines.

  • The statement does not call for a pause, but instead asks for an international effort to give governments and labs the choice to slow down if needed.

  • The letter pins the risk on automated AI research, warning models that improve other models could accelerate “beyond our ability to understand or control.”

  • Both OAI and Anthropic endorsed the letter on X, with OAI saying it believes “the world will need to pace the rate of AI advancement” in the future.

Why it matters: The AI era has produced a pile of open letters, but the loudest ones came from outside academics, safety nonprofits, or employees already out the door. This time feels like a vibe shift, with the calls now coming publicly from inside the frontier houses, who see the race nearing a point of no return.

Google turns Gemini into a Wispr Flow rival

As more people see the advantages of talking to their computers naturally using tools like Wispr Flow, Google has taken notice.

On Wednesday, Google launched an update to the Gemini app for macOS that allows users to long-press the Fn key to enable intelligent dictation. This lets users speak naturally into any desktop window and get polished text that removes pauses, “ums,” and even mid-sentence corrections.

Wispr Flow has offered this capability to macOS users since its launch in 2024 and has gradually expanded to Windows, iPhone and Android users due to high demand. The company has achieved such success that it is now valued at approximately $2 billion, and downloads are soaring. The promise it gives users is the ability to dictate more quickly than they can type, helping them move faster and, as a result, improve productivity. The average person can speak about 150 words per minute and type about 50 words per minute.

Google has already attempted to bring the Wispr Flow experience natively to its users via its Rambler feature for the Google Keyboard app (Gboard), announced in May and coming soon as an update on Android devices. It also uses AI to turn natural, unscripted speech into polished text. This desktop app upgrade, however, promises to offer users even greater productivity gains, as the laptop is where most working professionals, students, and other power users get their work done.

Beyond dictation, the Gemini app got another upgrade: on-screen reasoning. When users opt into Gemini reasoning in settings, Gemini understands the context on your screen to help with tasks. For instance, if you already have a document open on your screen, you can ask Gemini to take action on it without having to enter it manually or re-explain it, again helping you work faster.

The new voice capability is rolling out to the Gemini desktop app for macOS users globally in English, with more languages coming soon. While the company has yet to comment on when it will come to Windows users, it is common practice to first release to macOS and expand access in the following months.

What Else Happened in AI on July 30th 2026?

Moonshot AI reportedly raised $3.5B at a $35B valuation on the heels of its Kimi K3 release, with a potential Hong Kong IPO coming later this year.

OpenAI debuted a program offering 100K academic researchers free ChatGPT, noting that arXiv math papers crediting ChatGPT rose from 14 in February to 100 in July.

xAI released Grok Voice Think Fast 2.0, a speech-to-speech model that can think while it talks, beating top rivals on quality while cutting response delay to 0.7 seconds.

Lilian Wang reportedly re-joined OpenAI, coming just days after the Thinking Machines co-founder announced her departure due to stress and health challenges.

Google rolled out Lyria 3.5, the latest version of its music model, inside Flow Music, with upgrades aimed at more human-sounding vocals and control.

Elon Musk revealed that SpaceXAI’s next Grok 4.6 model is set to launch on Aug. 7, with 4.7 following just weeks later, which is ‘better than 4.6 in every way’.

OpenAI and Anthropic have reportedly been pressing Washington to subject their rivals to model reviews as well, coming ahead of Trump’s Aug. 1 framework deadline.

Recursive Intelligence partnered with AWS in a multi-year, $410M cloud deal to scale its self-improving system, coming two months after launching from stealth.

Anthropic showed that its Mythos Preview spent 60 hours halving the strength of a quantum-resistant algorithm, though it doesn’t threaten deployed systems.

Ant Group released Ling-3.0-Flash, a model for powering AI agents that it says rivals systems 2-3x its size, with the weights set to go open-source next week.

← All articles