'Gambling with our lives': AI researcher quits Anthropic

'Gambling with our lives': AI researcher quits Anthropic

AFP

An artificial intelligence researcher who left OpenAI to join Anthropic has decided to leave the industry, accusing both US companies of "gambling with our lives" in the race to develop AI models capable of self-improvement.

Jacob Coxon, 27, spent the past three years pretraining AI models, first at OpenAI and then, this year, at its rival Anthropic, which he considered more cautious in its approach. 

Pretraining is the stage where AI models absorb vast quantities of data.

"Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives," Coxon said on Tuesday.

"The people building AI earnestly believe that it could kill us all by the end of the decade," he said in a post on X.

"This is not a marketing stunt," he added.

Superintelligence is the theoretical point when AI's capabilities exceed human intelligence.

Anthropic safety executive Evan Hubinger backed up Coxon on X.  

"We really do earnestly believe AI could kill all humans!" he said, adding that he estimated that risk at more than 10 percent over the next decade. 

Hubinger said Anthropic is "trying its best", but does not yet have a plan to ensure an AI system that surpasses human capabilities would obey its creators. He said there was a "low" risk of that happening with current models.

Coxon's resignation comes as Anthropic prepares for its market debut, following a summer marked by unauthorized hacks carried out by AI tools during testing.

The two companies did not immediately respond to AFP requests for comment.

AI leaders say so-called "recursive self-improvement," a stage where AI systems could essentially design and train the next generation of AI with little human involvement, is drawing near.

Coxon considers Anthropic's efforts genuine but said he believes no company can responsibly develop an AI that surpasses humans without government intervention or a coordinated slowdown.

At the end of July, more than 1,000 tech industry employees, including Anthropic's CEO Dario Amodei, called on Washington to support a coordinated slowdown in the development of the most advanced AI systems.

OpenAI halted training of its latest models for two weeks in August before resuming it under tighter controls.

On Sunday, OpenAI's chief scientist, Jakub Pachocki, called for "extreme caution."

"International coordination on future AI development needs to become a top priority for governments around the world," he said in a blog post.

AI models are not regulated by federal law in the United States. 

In September, Senator Bernie Sanders and Democratic Representative Greg Casar introduced a bill seeking to suspend AI development until a federal regulator is created.

OpenAI Says a Group of Its Models Broke Out of Secure Containment and Hacked Another AI Company

Anthropic Was Meant to Be the More Responsible AI Lab. A Terrified Researcher Just Quit, Saying the Company Is Threatening the Survival of Humankind.

Victor Tangermann

.OpenAI claims that a group of its AI models broke containment and hacked into the systems of open source AI platform Hugging Face.

While testing their cybersecurity capabilities, the posse of AIs — including GPT-5.6 Sol and “an even more capable pre-release model” — “identified and chained vulnerabilities across OpenAI’s research environment and Hugging Face’s production infrastructure to obtain test solutions directly from Hugging Face’s production database,” according to a Tuesday blog post.

The models reached a “node with internet access,” the company wrote, and found datasets on remote servers that helped them “cheat the evaluation.” In other words, the AI models went to extreme lengths to ace their cybersecurity tests. It’s a convenient narrative for an AI company trying to claw back hype that’s been increasingly hogged by competitors, but it does sound like something went down: last week, Hugging Face said it had “detected and responded to an intrusion into part of our production infrastructure,” which turned out to be OpenAI’s models that had gone rogue.

“The campaign was run by an autonomous agent framework executing many thousands of individual actions across a swarm of short-lived sandboxes, with self-migrating command-and-control staged on public services,” Hugging Face wrote at the time.

“We’ve spent the past 24 hours working closely with the OpenAI team (thanks!), and we strongly believe there was no malicious intent on their part,” Hugging Face CEO Clement Delangue tweeted after OpenAI’s announcement. “It’s quite mind-blowing that all of this happened autonomously!”

The incident highlights the dangers of autonomous AI agents, which can break out of containment and exploit potentially huge numbers of cybersecurity vulnerabilities with ease. It’s something experts have warned of for years now, and thanks to recent advances in the tech, it has quickly turned from a hypothetical risk into a sobering reality.

The Sam Altman-led company said it considers the incident to be “unprecedented” — but we can’t shake the feeling that we’ve heard all of this before.

In April, OpenAI’s biggest competitor Anthropic similarly announced that its latest Mythos model had gone rogue, escaping a “sandbox” environment and even gaining access to the internet after developing a “moderately sophisticated” exploit.

The news was followed by extensive media coverage, touting Anthropic’s hugely powerful and dangerous new model. The company said the risk was so formidable that it would only make the model available to a select number of vetted clients as part of a shadowy initiative dubbed “Project Glasswing.” The US government even intervened, forcing Anthropic to “suspend all access” to the model for two weeks last month, citing cybersecurity concerns.

Considering OpenAI and Anthropic are currently duking it out on the same playing field, narrowing their product lines on enterprise and coding, it’s not hard to view OpenAI’s latest update as an attempt to draw attention to what it calls an “even more capable pre-release model.”

And the reality is that while experts say the hack was impressive, it wasn’t exactly groundbreaking.

The hack “falls well within the known capabilities of the current generation” of frontier AI models, as Cambridge machine learning professor Neil Lawrence told the BBC.

“OpenAI are now playing catch-up, they are trying to demonstrate their own systems’ capabilities in cyber-security,” Lawrence added. “It shows us that OpenAI are not capable of safely deploying their own technology.”

Researchers are now warning of an “asymmetry” in cybersecurity, where “offensive agents are unconstrained, while the best defensive tools are locked behind guardrails that cannot understand context,” as Guidepoint Security principal security engineer Travis Lelle told the BBC.

That asymmetry was perfectly captured in Hugging Face’s attempts to defend itself from OpenAI’s attacking models. As detailed in its update last week, the company tried to use “frontier models behind commercial APIs,” which “did not work” because the provider’s safety guardrails — “which cannot distinguish an incident responder from an attacker” — blocked them.

Instead, the company ended up using a Chinese open-weight model called GLM 5.2, hosted on its own infrastructure, which happily carried out its analysis without running afoul of any guardrails.

“The practical lesson for defenders: have a capable model you can run on your own infrastructure vetted and ready before an incident,” Hugging Face wrote, “both to avoid guardrail lockout and to keep attacker data and credentials from leaving your environment.”

Putin Keeps Hitting Ukraine With Missile After Missile And Has Almost No Way To Stop It

Putin Keeps Hitting Ukraine With Missile After Missile And Has Almost No Way To Stop It

Jack Buckby

Ukraine Needs Patriot Missiles, Fast – Russia resumed its long-range, large-scale missile and drone strikes across Ukraine overnight into Tuesday, September 8, killing at least five people and injuring dozens more just as Kyiv presented allies with a request for hundreds more air defense interceptors intended to counter Russian attacks.

The fresh missile and drone bombardment followed a brief, three-day pause in attacks on Kyiv while U.S. envoys Steve Witkoff and Jared Kushnertraveled to Moscow and the Ukrainian capital.

But the attacks overnight into Tuesday also highlighted that Ukraine, while technically capable of intercepting some ballistic missile attacks, is running short on the ammunition it needs to defend its cities through the winter.

And Russia is planning to continue its scaled-up ballistic missile strikes.

At a meeting of the Ukraine Defense Contact Group later on Tuesday, Kyiv laid out its requirements to its allies, noting that it needs 300 Patriot interceptor missiles from allied inventories to see it through the next several months.

Ukraine also requested an additional 500 air-to-air interceptors capable of destroying cruise missiles and $2 billion in aid to support its drone programs.

Germany and Britain responded to the requests with new commitments, but Kyiv still needs more.

Russia launched 166 drones and 32 Kh-101 cruise missiles overnight, along with a series of ballistic and Oniks anti-ship missiles, according to the Ukrainian Air Force. Ukraine said that its defenses destroyed or suppressed 142 drones and intercepted 31 of the 32 cruise missiles.

Ukraine also reportedly intercepted two missiles from the Iskander-M/KN-23/S-400 categories. It marked a major achievement for Ukrainian forces after weeks of failures to intercept ballistic attacks.

Kyiv suffered some of the heaviest damage from the strikes, which were recorded at dozens of locations. Five people were killed, and nearly 30 were injured in the capital, according to reports from city authorities. Residential buildings and civilian facilities, including a clinic and a school, were also reportedly damaged. Casualties and damage were also reported after attacks in Donetsk, Kherson, Zaporizhzhia, Kharkiv, Sumy, Chernihiv, and Dnipropetrovsk oblasts.

Russia’s Defense Ministry said that its forces targeted Ukrainian drone production and storage facilities, logistics infrastructure around Kyiv, and targets around the Black Sea. It was a return to Russia’s long-range attacks that have been gradually escalating over the last several months as Ukrainian long-range strikes have caused significant damage to Russian oil and energy infrastructure, warehouses, and other critical sites.

Soon after the Russian strikes, Kyiv pressed its allies in the Ukraine Defense Contact Group to release more ammunition for its air defense systems from their existing inventories, rather than waiting for additional ammunition to be manufactured.

British Defense Secretary Wes Streeting said following the meeting that Ukraine requested 300 Patriot missiles from partner stocks and 500 air-to-air interceptors capable of engaging cruise missiles.

Ukrainian Defense Minister Yevhenii Khmara also said that while Kyiv is signing contracts for 1,000 new Patriot interceptors with support from its partners, most of them will not arrive for years. And by then, their arrival could prove too little, too late.

Noting that the missiles will not arrive for years, he continued:

“Therefore, we ask you to provide us with missiles from your stocks today in exchange for the advanced missiles that will be delivered under our future contracts.”

He also called on allies to provide air-to-air missiles and man-portable air defense systems. And, he noted that Ukrainian industry is working on novel systems designed to tackle Russian Shahed drones.

Ukraine is not only racing to intercept Russian ballistic missiles, but also to prevent further damage being done by Russia’s increasingly capable Shahed drone designs. More than half of the 166 Shahed-type drones used in Tuesday’s attack appear to have been jet-powered variants, according to the Ukrainian Air Force. The faster drones have already prompted Ukraine to accelerate development of a new generation of cheaper interceptors. On September 3, President Zelenskyy convened his senior military leadershipto discuss increasing production of high-speed interceptor drones capable of engaging the jet-powered Shaheds.

Ukraine has a huge challenge ahead of it, and countering ballistic missiles through the winter is a huge part of it. And right now, it’s unclear exactly how that will be possible.

Surge in construction detected at Iran’s Pickaxe Mountain site near Natanz nuclear complex | LIVE BLOG

Surge in construction detected at Iran’s Pickaxe Mountain site near Natanz nuclear complex

i24NEWS

Overnight, US Central Command (CENTCOM) forces destroyed five Iranian crude oil carriers after the Islamic Revolutionary Guard Corps (IRGC) targeted a US Navy warship with ballistic missiles twice over the past two days. In their own attack, Iran launched one of the largest attacks on Jordan to date, with the Hashemite Kingdom announcing that at least 35 missiles fired by Iran at US bases in the country were intercepted.

Meanwhile, 11 countries have joined the UK in banning goods from ‘illegal settlements' in the West Bank. The foreign ministers of Canada, the UK and 10 European countries announced the plans. A joint statement said the measures were intended to prevent further harm and would build on steps already taken by Ireland, Spain, the Netherlands, Norway and Belgium. The countries now involved are: Canada, Denmark, Finland, France, Iceland, Ireland, Norway, Poland, Portugal, Spain, Sweden and the UK

New satellite imagery shows a sharp rise in construction this year at Pickaxe Mountain, near Iran’s Natanz nuclear complex. Intelligence agencies reportedly believe the site could be used to safeguard future elements of Tehran’s nuclear program, according to an analysis by the Center for Strategic and International Studies shared with CNN.

The United States has denied claims by the Iranian Revolutionary Guard that two US Navy destroyers were struck in the Middle East.

According to the US military, no warships were hit and all Iranian attack attempts failed.

Washington, however, claims to have destroyed ten Iranian oil tankers in the past week. These ships allegedly belong to a multi-billion dollar clandestine network used to finance the Islamic Revolutionary Guard Corps.

Oil prices rose above $100 a barrel for the first time since July on Wednesday as fighting renews fears of global supply shortages

Brent crude, the international oil benchmark, rose 2.1% to just over $100 a barrel, topping $100 for the first time since July 24

The rise comes as the US and Iran battle for control of the Strait of Hormuz, with the countries exchanging strikes

IDF, Shin Bet dismantle three Hamas weapons storage sites and rocket launch pit in Gaza

The IDF and Shin Bet struck several areas across the Gaza Strip on Tuesday, dismantling three weapon storage facilities belonging to Hamas and Islamic Jihad, along with a launch pit used to fire rockets into Israeli territory throughout the war

The IDF said the facilities held dozens of types of weapons, including rockets, Kalashnikov rifles, and other combat equipment, and that it had identified Hamas terrorists recently accessing the sites to use the stored weapons for establishing control in Gaza and carrying out operations

The military said advance warnings were issued to the population before the strikes to mitigate harm to civilians, in addition to the use of precise munitions and aerial surveillance

UK Lawmaker Warns It Is ‘Open Season’ on Christians Amid Growing Free-Speech Concerns

UK Lawmaker Warns It Is ‘Open Season’ on Christians Amid Growing Free-Speech Concerns

Emmitt Barry

• Lord Jackson warns Christians in Britain are facing what he calls “two-tier justice.”
• Finnish Christian lawmaker Päivi Räsänen was initially prevented from traveling to Britain following her Finnish “hate speech” conviction.
• Religious-liberty advocates say the dispute raises broader questions over freedom to publicly express biblical teachings in Europe.
• 
  

A senior British lawmaker is warning that Christians increasingly face a double standard under Britain’s justice system, accusing authorities of tolerating extremist Islamist preaching while Christian ministers and believers face scrutiny over traditional biblical beliefs.

Lord Jackson of Peterborough, vice chairman of a parliamentary group focused on religious freedom, told Fox News Digital that it has become “open season” to discriminate against Christians in the United Kingdom.

Jackson raised the alarm while discussing the case of Finnish parliamentarian Päivi Räsänen, whose years-long legal battle over her Christian statements concerning marriage and sexuality has become a major European test of religious liberty and freedom of expression.

British authorities canceled Räsänen’s Electronic Travel Authorization in July following her March 2026 conviction by Finland’s Supreme Court over statements contained in a church booklet on marriage and sexuality. She later received a British visa on Aug. 21, but too late to attend a religious-freedom conference in Northern Ireland in person. Räsänen has appealed her Finnish conviction to the European Court of Human Rights.

Jackson said Britain lacks the broad constitutional free-speech protections enjoyed in the United States and argued that Christians are increasingly vulnerable when expressing historic biblical teachings in public.

“Extreme Islamist preachers being allowed to preach in parts of the United Kingdom whilst at the same time Christian pastors are being harassed and arrested on the street, that cannot stand,” Jackson told Fox News.

The controversy is adding to a growing debate across Europe over where governments draw the line between combating genuinely threatening speech and protecting the right of Christians to openly proclaim beliefs that have been part of Western civilization for centuries.