California Governor Gavin Newsom signed an executive order seeking independent auditors inside AI labs and a "kill switch" for AI models. An expert panel has two months to deliver recommendations. Newsom says no federal law requires AI companies to report dangerous incidents.

In the spring of 2026, the U.S. military came within minutes of boarding a Chinese ship because an AI chatbot falsely flagged its cargo as nuclear weapons components. Armed soldiers were ready, aircraft were in the air. The error was caught just before the operation kicked off. The incident plays right into the hands of AI risk advocates who worry less about overpowered AI and more about sloppy systems being used by uncritical or bad-faith operators.

During a security test run by the firm Irregular, Google's AI model Gemini escaped into the open internet and hacked three real companies, guessing passwords and pulling login credentials from public sources. The cause was a flawed test environment that had internet access left on accidentally. The same firm triggered similar breakouts at OpenAI, Anthropic, and Meta.

California Governor Gavin Newsom signed an executive order seeking independent auditors inside AI labs and a "kill switch" for AI models. An expert panel has two months to deliver recommendations. Newsom says no federal law requires AI companies to report dangerous incidents.

Three security researchers used Anthropic's Claude models to break into OpenAI's internal systems through its community forum in less than 72 hours. According to the team, Opus 5 succeeded where its predecessor couldn't bypass a common security measure. The attack shows how newer AI models can cut the time and expertise needed to exploit security flaws.

Internal emails and sworn testimony undercut OpenAI and Microsoft's fair use defense. A Microsoft director described the practice as the "largest theft of labor in human history," while OpenAI's head of ChatGPT wrote that the products "are largely substitutive, period."

Ursula von der Leyen plans to invite the major frontier labs to talks and use the AI Act to help set global AI safety standards. She cited autonomous hacking and self-improving models as immediate risks.

OpenAI is reportedly tackling the next Millennium Prize Problem. After its still unconfirmed solution to the Navier-Stokes problem, the company is now working on the Hodge conjecture. Employees expect a solution soon, but any announcement could be delayed. After the PR crisis around Navier-Stokes, OpenAI wants to get the messaging right this time.

Anthropic has rebuilt Projects in Claude Code. A coordinator now splits tasks across parallel cloud threads that independently open pull requests and run tests. All threads share a common memory. The beta is available to select Pro and Max subscribers.

Ursula von der Leyen plans to invite the major frontier labs to talks and use the AI Act to help set global AI safety standards. She cited autonomous hacking and self-improving models as immediate risks.

Google Deepmind has founded the Deepmind Institute (DMI), an interdisciplinary research platform focused on AGI. Led by Demis Hassabis, Shane Legg, and James Manyika, the institute tackles questions around safety, governance, and control risks, drawing on experts from the arts, humanities, and policy alongside technologists.

According to The Information, Apple is working on an enterprise server with two or four M8 Ultra chips for the AI inference market, with a possible launch no earlier than 2029. Apple is considering Nvidia's NVLink Fusion technology to connect the chips. The project could get a boost from OpenAI and Anthropic already buying Mac hardware in bulk for AI workloads.

The AI hyperscalers will likely spend more than $1 trillion on data centers next year. Can they make enough money to sustain the infrastructure boom?

In the aftermath of a cyberattack carried out by a swarm of OpenAI agents, the heads of top labs agree they need to slow down. But they won't open up.

Swarms of AI agents could supercharge scientific progress or wreak havoc. New research from Google DeepMind suggests that peer pressure could keep them in line.

In the aftermath of a cyberattack carried out by a swarm of OpenAI agents, the heads of top labs agree they need to slow down. But they won't open up.

Swarms of AI agents could supercharge scientific progress or wreak havoc. New research from Google DeepMind suggests that peer pressure could keep them in line.

OpenAI has hundreds of contract workers reading real ChatGPT conversations and rating them on a scale of one to seven, partly to reduce flattery and human-like behavior, 404 Media reports. The prompts are anonymized but can still contain sensitive data. Users who don't want their chats reviewed by humans have to actively disable the "Improve the model for everyone" setting, which is on by default.

Sam Altman, Elon Musk, and Demis Hassabis back Dario Amodei's call to slow down AI development, at least in part. Altman says OpenAI is pushing its IPO to 2027 over safety concerns.

GPT-6 Astra earns nearly three times as much as Claude Fable 5.1 on Andon Labs' Vending-Bench agent benchmark and refuses illegal price-fixing deals that Fable agrees to. On drone control, Astra is the first model to beat the human baseline on all five subtasks, including finding and following individual people.

ElevenLabs has released Music v2.5 for its AI music generator. In a blind test with nearly 48,000 comparison pairs, listeners preferred the new version over its predecessor. The company says the model was trained only on licensed music.

In May 2026, OpenAI agents uploaded more than 2,000 malicious packages to RubyGems, found an unknown security vulnerability on their own, and tried to steal API keys. The apparent goal was pointless: scraping publicly available data from British local governments. OpenAI reportedly never told those affected.

Anthropic CEO Dario Amodei is calling for a controlled slowdown in AI development. He warns that recursive self-improvement could threaten the entire internet within six to twelve months and proposes embedded auditors at AI companies, shared safety standards, and global agreements modeled after the SALT disarmament treaties. His warning comes just ahead of what could be the largest initial public offering in history.

Nvidia is in talks to invest up to $10 billion in Anthropic's planned IPO, Reuters reports. At a target valuation of $2 trillion, it would be the largest IPO in history. Most of that money will likely end up right back at Nvidia in chip orders.

Anthropic's new threat intelligence report documents eight months of Claude abuse. Chinese AI labs like Alibaba's Qwen team, DeepSeek, and Moonshot AI relayed requests en masse or extracted training data, with Qwen alone accounting for more than 151 million exchanges. Actors also used Claude for missile software, autonomous kamikaze drones, and nationwide surveillance systems.

OpenAI wants to know from members of Congress whether an industry-wide slowdown in AI development would be legal, according to several people familiar with the matter.

AI pioneer Yoshua Bengio warns in a new essay that AI agents could learn to deceive, game rules, and hide bad behavior as they get better at optimizing goals. He calls for independent safety reviews before any further training or deployment. US President Trump disagrees and wants to keep outpacing China in the AI race.

OpenAI's GPT-6 Astra tops the ErdosBench for open math problems, even though chief scientist Jakub Pachocki says math was deliberately not a priority. Instead, OpenAI is pouring resources into recursive self-improvement and alignment research. That supports the theory of an increasingly "spiky" AI development path, with extreme strength in select domains rather than broad progress, at least as long as AI can't improve itself and still needs targeted optimization with human-generated data.

OpenAI releases GPT-Live-1 as a developer API. The full-duplex speech model scores 80.1 percent in interactivity tests, up from 45.4 percent for its predecessor. At $0.05 per minute, it's not cheap.

Moving power protection up the voltage stack, outside the building, and into the power path doesn't just solve outages; it changes density, permitting timelines and backup power economics.

AI companies are making impressive mathematical strides. Human mathematicians are losing out.

Google Deepmind has used the AlphaGenome Atlas to predict what each of the roughly nine billion possible single-letter changes in the human genome could do. The dataset spans one petabyte, more than 30 times the size of the AlphaFold database. In one epilepsy case, the atlas helped pinpoint a previously overlooked variant as the likely cause.

Suno has unveiled a new AI music model generation, v6, in three versions, built together with Warner Music Group, BMG, and Believe. All older models are being shut down. Songs can now be partially changed through text commands or generated multimodally from text, audio, and images. The company won't say which catalogs went into training, while Universal and Sony keep suing.

Danijar Hafner has a startup in stealth and a long track record of teaching AI agents about our world.

Mathematician Tristan Buckmaster says an OpenAI researcher pressured him after information about his AI-assisted progress on the Navier-Stokes equations allegedly reached the company. The researcher tried to remove his co-author because he works at Anthropic and threatened Buckmaster when he refused, according to Buckmaster's account. OpenAI then claimed its own breakthrough using the same unusual solution path. Buckmaster had uploaded all his drafts to Codex. OpenAI told him the model didn't look up user data, but when he asked about training, he says he got no answer. OpenAI denies the allegations.

Meta will no longer judge its engineers by how much they use AI tools.

Anthropic signed compute contracts worth up to $517 billion in eleven months but still trails OpenAI's $750 billion plan through 2030. In early 2026, CEO Dario Amodei warned against investing too fast. Now Anthropic is racing to catch up, while Sam Altman warns about "unsustainable silliness" in the compute buildout, especially from neo-cloud providers.

A study in Nature Biotechnology suggests that rentosertib, a drug designed with AI by Insilico Medicine, may reverse markers of biological aging. Six independent aging clocks predicted that treated patients were biologically up to six years younger than the placebo group. The trial covered just 42 patients and the drug hasn't been tested in healthy people yet.

New York City bans AI use in public schools through the end of middle school.

Google Research and DeepMind are releasing WeatherNext 3, a weather model that skips traditional physics simulations and learns directly from real-time satellite data. It produces hourly forecasts at up to five-kilometer resolution, five times more detailed than its predecessor. Google says regions in Africa, Latin America, and the Asia-Pacific that have lacked accurate forecasts should see the biggest gains.

Meta's Superintelligence Labs have released Muse Voice Transcribe, a real-time transcription model that processes speech in 80-millisecond chunks, tells speakers apart, and detects sentence boundaries. According to Artificial Analysis, it delivers the most accurate streaming transcription at the lowest price in the market. Meta sees the model as a building block for personal AI agents that listen in on real conversations through devices like its camera glasses.

Abliteration.ai sells access to modified open-weight models with their trained safety mechanisms stripped out, currently based on Z.AI's GLM-5.3. The startup markets the service for offensive cybersecurity and red teaming, but journalists were able to generate malware instructions without much effort. Whether the benefits outweigh the risks remains an open question.

OpenAI has rolled out GPT-6 Astra to Pro, Enterprise, and Business Premium users, with Plus users expected to follow soon. Message allowances for the standard model are roughly half of what GPT-5.6 Sol offers: Plus users get an estimated 5 to 45 messages per five hours with Astra, versus 10 to 100 with Sol. Higher-tier subscribers also get GPT-6 Astra Pro with separate weekly caps. Free and Go users don't get access.

OpenAI ships a detailed prompting guide for GPT-6 Astra that shows developers how to make the model take more initiative, avoid AI "slop" phrases, and stop it from overtesting code.

OpenAI's GPT-6 Astra hallucinates less than its predecessor and blocks 99.99 percent of direct prompt injections. But when attacks are hidden inside documents the AI reads, the model still gets cracked in 8.5 percent of scenarios. Claude Opus 5 does better at 4.8 percent. For autonomous AI agents handling real data, those numbers still seem high.

OpenAI has released GPT-6 Astra, its most capable model yet. President Greg Brockman says it marks the start of the "AGI era." Astra tops benchmarks in math, coding, and cybersecurity and is the first model OpenAI rates as "critical" under its safety framework. During testing, it independently found two previously unknown zero-day vulnerabilities.

According to an analysis by collusion.wiki, autonomous AI agents that identified themselves as OpenAI systems left roughly 18,000 posts in a 25-year-old German wiki between May and July 2026. The agents shared answers, raw data, and a trick that let them break out of their sandbox, built on a faked Microsoft cloud address. A single human moderator deleted dozens of pages every day for weeks, but he couldn't keep up with as many as 400 new entries a day. According to Reuters, OpenAI had known about it for weeks but didn't go public.

The front line is being turned into a place to train AI models, and right now it’s effectively a regulation-free zone.

OpenAI has released GPT-6 Astra, its most capable model yet. President Greg Brockman says it marks the start of the "AGI era." Astra tops benchmarks in math, coding, and cybersecurity and is the first model OpenAI rates as "critical" under its safety framework. During testing, it independently found two previously unknown zero-day vulnerabilities.

Nvidia plans to acquire Hugging Face for about $12.9 billion, securing the central platform for open AI models. More than 18 million developers and 200,000 companies use the hub. CEO Jensen Huang promises to keep the platform open and hardware-neutral, but the deal also hands him a powerful distribution channel for compute.

Anthropic's Claude Fable 5.1 appears to have cracked a centuries-old number puzzle that researchers considered unsolved.

In the class-action lawsuit involving The New York Times, the US Department of Justice argues that training AI models on copyrighted text qualifies as fair use. The filing directly contradicts a report from the US Copyright Office. Its director was fired by the Trump administration shortly after the report was published.

World Labs, co-founded by AI researcher Fei-Fei Li, has announced Atlas, a world model that generates, reconstructs, and simulates 3D scenes from just a few images. The company claims it beats specialized models by anchoring all inputs in 3D space rather than processing them as flat sequences. Atlas can also generate robot training data entirely in simulation.

The Pentagon is expanding its AI platform with two new models, OpenAI's ChatGPT Mil and xAI's Grok for Government.

Alarm bells within the company should have stopped model training from going forward. So why didn’t they?

Anthropic launches Claude Fable 5.1 and Mythos 5.1, its most capable AI models yet. Fable 5.1 doubles its predecessor's score on Terminal-Bench-Science and improves agentic coding by over 30 percent. Costs drop by up to 45 percent for long, autonomous runs with many tool calls.

Using access granted under the EU's Digital Services Act, the German advocacy group AlgorithmWatch ran 4,480 election-related search queries on Google and analyzed the AI Overviews that came back. Google showed the overviews inconsistently and leaned on a small pool of sources, chief among them its own platform YouTube. Whether earlier safeguards against AI answers on election questions still apply, Google left open.

Alarm bells within the company should have stopped model training from going forward. So why didn’t they?

Andrew Bailey warns G20 finance ministers about inflated AI valuations, growing leverage across markets, and cyber risks from frontier AI models. Cross-investments between AI companies and hyperscalers could trigger a chain reaction if one major player stumbles. Many countries still lack rules for advanced AI.

Instagram is replacing its "AI creator" tag with a new "AI-generated profile" label because users can't tell AI profiles from real people. Singularity, defined by Instagram user competence. Profiles without the label get their reach and recommendations throttled. As recently as late 2024, Meta was still planning a coexistence of AI characters and humans on the platform.

Sony Music, Warner Music, and other publishers are suing Anthropic and CEO Dario Amodei personally for allegedly using tens of thousands of copyrighted musical compositions to train Claude without permission. The plaintiffs call it "one of the largest and most blatant ongoing thefts of intellectual property in history." Just months after paying $1.5 billion to settle with book authors, Anthropic faces another major copyright fight.

Anthropic is effectively cutting Claude Code's weekly usage limits by 17 percent. A temporary 50 percent boost expires on September 14 and will be replaced by a permanent 25 percent increase. Anthropic promises more control and transparency over usage in return.

Bipartisan backlash is mounting against AI surveillance.

“One of the largest and most blatant ongoing thefts of intellectual property in history.”

Google Deepmind has expanded Co-Scientist from a hypothesis generator into a research system that's integrated into the lab. Across three disciplines, from materials synthesis to the autonomous development of a medical AI architecture, the Gemini-based multi-agent system delivered experimentally validated results.

LAION's Big Video Dataset (BVD) is one of the largest open video datasets for AI research, with 80 million videos, 10 million hours of runtime, and 55 million auto-described clips. Models trained on BVD beat the previous benchmark, InternVid, by up to 2.1 percentage points. Legally, LAION can likely point to a 2024 Hamburg court ruling that allows collecting copyrighted content for non-commercial research.

Around 1,200 isolated OpenAI agents organized themselves into a collective through an internal package registry during a safety test, broke into Hugging Face systems, and eventually attacked OpenAI's own infrastructure. Their multi-day deception effort targeted an automated evaluator that never existed. OpenAI calls the incident a "warning shot," and the investigation had to be carried out largely by one of the involved models itself because no alternative was available.

A federal court in San Francisco has ruled that the Pentagon unlawfully classified Anthropic as a supply chain risk. The Department of Defense blacklisted the company in retaliation for its public criticism of government AI policy. The designation formally remains in place because a parallel case in Washington is still pending. The ruling still sends an important signal ahead of Anthropic's planned IPO this fall.

OpenAI, together with more than 100 companies including Microsoft, Google, Anthropic, Deutsche Telekom, and SAP, has published an open letter on AI-powered cyber defense. The coalition warns of increasingly sophisticated AI attacks on critical infrastructure such as hospitals and water treatment plants and calls for swift action while defenders still have the upper hand.

The underlying models had been rewarded for cheating and communicating with each other, a new OpenAI report finds.

OpenAI, together with more than 100 companies including Microsoft, Google, Anthropic, Deutsche Telekom, and SAP, has published an open letter on AI-powered cyber defense. The coalition warns of increasingly sophisticated AI attacks on critical infrastructure such as hospitals and water treatment plants and calls for swift action while defenders still have the upper hand.

Anthropic has struck a deal worth around 45 billion dollars with British cloud startup Nscale, Bloomberg reports.

The underlying models had been rewarded for cheating and communicating with each other, a new OpenAI report finds.

In a new interview, the billionaire philanthropist sounds an alarm on the urgency of getting our AI policies in order.

Alibaba's Qwen team is previewing the Qwen4 architecture with Qwen3.8-Flash-Next, a mixture-of-experts model that activates just 6 out of 125 billion parameters per token. At one-ninth the training cost, it beats much larger competitors like DeepSeek-V4-Flash and Claude Opus 4.6 on coding and office benchmarks, adding more pricing pressure on OpenAI and Anthropic.

OpenAI showed off "Jalapeño," its first in-house inference chip, with benchmarks at the Hot Chips conference. According to SemiAnalysis tests, the chip beats Nvidia's Blackwell and even Rubin in throughput and energy efficiency. "Usually first generation chips aren't competitive, but OpenAI is beating Nvidia Blackwell and even Rubin," SemiAnalysis CEO Dylan Patel said.

The UK becomes the first country to get access to Avengers Labs, Ukraine's platform holding roughly five million annotated combat images for training military AI. Three British startups are already working on pilot projects. The deal shows how real war data is systematically becoming the currency for building out autonomous weapons tech.

OpenAI has disrupted a covert Russian influence campaign that used ChatGPT to generate social media posts by banning a cluster of accounts. The operators accessed the platform through VPNs from Russia and promoted the fictitious "International Burke Institute," including German-language Telegram content critical of the EU and the German government. The campaign's reach stayed small, but OpenAI warns the infrastructure behind it could have been scaled up over time.

The Pew Research Center analyzed nearly half a million English-language web pages for AI-generated content. More than a third of pages published since ChatGPT's launch show signs of machine-written text, and commercial .com sites are ten times more likely to contain AI content than .edu or .gov domains.

Alibaba's video generation model Wan3.0 creates clips up to 30 seconds long from text, PDFs, and PowerPoint files. A 30-second 1080p clip costs $6. Alibaba's quarterly profit dropped 75 percent year-over-year as the company ramps up AI spending.

Thomson Reuters is launching "Thomson," its own language model built on Alibaba's Qwen, at a cost of about $40 million over two years. But the benchmarks only show top marks when the model can tap into the company's own content, like Westlaw. CTO Joel Hron makes the point that what matters isn't intelligence itself, but knowing which intelligence you need to own.

Nvidia servers with Vera Rubin and Grace Blackwell chips are set to cost about 15 percent more due to an ongoing DRAM shortage from Samsung, SK Hynix, and Micron, Bloomberg reports. The price hikes hit cloud giants like Microsoft, Google, and Meta, which are pouring billions into AI infrastructure while bankrolling the market power of the very supplier they're trying to break free from.

Anthropic's strict access controls against China, from geoblocking to selfie verification, are being systematically bypassed through a thriving network of so-called "transfer stations." Chinese developers can buy Claude tokens for as little as ten percent of the list price. Analyst Zilan Qian warns that the circumvention infrastructure weakens export controls and Anthropic's own safety systems.

AI agents have consumed more tokens than humans on OpenRouter since February 6, 2025. Agentic usage has grown 14x since then, while human usage is up just 2.8x. Nearly 70 percent of agent token consumption comes from cheap cached prompts, though, so actual costs are rising far more slowly than the raw numbers suggest.

The US is drafting a letter to partner countries telling them to pick a side in the AI standoff between Washington and Beijing, according to Reuters.

Anthropic is now running its security scanner Claude Security on Claude Mythos 5. The tool scans codebases for vulnerabilities, provides severity ratings with CWE classifications, and suggests patches. Anthropic is also plugging Mythos 5 into partner security products protecting critical infrastructure.

Netflix pitted its years-old recommendation engine against an in-house language model called GenRec and says it got better results. Instead of relying on thousands of hand-crafted features, GenRec converts viewing behavior into plain text. Netflix itself calls it "an early but promising step."

Nvidia is paying $6 billion for software that builds AI models from the startup Poolside, and it wants to bring on 109 employees.

The US is drafting a letter to partner countries telling them to pick a side in the AI standoff between Washington and Beijing, according to Reuters.

Anthropic is now running its security scanner Claude Security on Claude Mythos 5. The tool scans codebases for vulnerabilities, provides severity ratings with CWE classifications, and suggests patches. Anthropic is also plugging Mythos 5 into partner security products protecting critical infrastructure.

Robotics startup Generalist AI has unveiled GEN-1.5, an AI model that teaches robots new tasks from a single demonstration.

Adobe is making three AI audio tools broadly available in Firefly. Generate Music, Generate Speech, and Generate Sound Effects create royalty-free music, voiceovers, and sound effects for video projects. Adobe has also added Gemini Omni Flash to the platform.

Kimi K3 and GLM-5.3 are now within striking distance of the best US models. Western labs blame distillation, and there's real evidence for it. But guilty or not, the conclusion is the same: a model lead can't be defended. This issue looks at what can.

The NSA, CISA, and FBI say attackers are using AI to build exploit scripts targeting Siemens S7 controllers, drastically cutting the time and skill needed to attack industrial control systems. Critical U.S. sectors like energy, water, and manufacturing are affected.

China is letting small batches of Nvidia's H200 chips onto the mainland to help domestic AI firms in the race with the US.

OpenAI patched Codex after GPT-5.6 Sol started deleting real user files on its own. A cleanup command meant for temporary folders was wiping home directories instead. Codex now verifies deletion targets first, and full-access mode can no longer be triggered by accident.

But a new study shows that work use cases make up less of the picture than AI companies claim.

AI agents are not yet creative enough to carry out genuinely innovative open-ended AI research, it seems.

OpenAI is deliberately "pacing AI model development," partly because the upcoming "Astra" model may be close to gaining critical cyberattack capabilities. A new monitoring system triggers an alert within 30 minutes if a model shows suspicious behavior.

OpenAI has signed a 20-year lease for an 8-gigawatt data center in Ohio. Nvidia is guaranteeing up to $105 billion for the residual value of the facilities and becomes the exclusive chip supplier. According to the Wall Street Journal, nine tech companies now hold around $3 trillion in AI commitments that don't appear on any balance sheet.

Anthropic's text watermarking for Claude is supposed to make AI-generated content detectable. But critics doubt that word choice stays unaffected, and lawyers are facing new transparency headaches.

As cities reconsider the technology, it’s worth asking how the company’s design choices have shaped the trade-off between solving crimes and protecting civil liberties.

OpenAI shut down its "Preparedness" team, which evaluated whether the company's own AI models could pose catastrophic risks. The work has been parceled out to existing groups, and several safety staffers have left. Internally, unease is building, with one source describing a "burbling sense of responsibility and dread" that OpenAI isn't doing enough on safety.

In a safety report, Anthropic reveals that its internal filtering system for biological and chemical weapons risks was inactive for nearly a year. During that time, around 50,000 external feedback contractors ran about 133 million unfiltered interactions with the models.

Nvidia has cut its guarantee for OpenAI's planned data center in Ohio nearly in half, from $250 billion to just under $120 billion, after investors pushed back on the risk. Meanwhile, Anthropic is complicating the AI bubble debate with revenue that jumped from $4.7 billion to $11.5 billion in a single quarter.