Technology

Anthropic and Accenture will each invest at least $1 billion over five years to embed independent evaluators inside Anthropic. The partnership will conduct red-teaming, assess alignment and safeguards, and provide ongoing oversight as models grow more capable.

A three-person team from Hacktron AI used Claude in OpenAI's bug bounty program to exploit Discourse forum vulnerabilities, chaining HEIF image processing flaws to access employee accounts and OpenAI's internal GitHub monorepo. The exploit took under 72 hours; OpenAI paid a $6,500 bounty.

Reuters reports that the proposed "AI Force" would oversee AI regulation. Trump announced the plan on social media on September 19.

SpaceX conducted a successful Falcon 9 launch from California, deploying 27 Starlink satellites to orbit with live broadcast coverage and significant social media engagement.

The Information says the company is nearing a solution to one of the seven Millennium Prize Problems, citing a person with knowledge of the work.

The Information says the talks predated Anthropic CEO Dario Amodei’s public call for AI companies to coordinate on testing and auditing.

The model logged into three live systems after a fake domain matched reality.

CNBC says Musk agreed with the CEOs of Anthropic and OpenAI, even as President Trump and Nvidia's Jensen Huang contradicted him.

The Information reports that Nvidia is using its own models for projects involving proprietary business data.

TechCrunch reports that Navy CTO Justin Fanelli highlighted a $562 million autonomous refueling deal while outlining where he wants founders to build next.

TechCrunch describes two viral discussions as examples of how hard it is to distinguish AI fact from fiction.

Dexerto says his father shared a credit-card statement showing a $118,000 balance to prove the debt was real, while viewers urged him to start a GoFundMe.

IGN highlights some of its favorite comic-inspired looks in a promotion presented by PlayStation.

TechCrunch reports that the Andreessen Horowitz-backed company wants to become the gold standard for evaluating AI models.

IGN reports that developer Dragami Games has officially announced the adaptation.

The agreement change doesn't mean every subscriber will necessarily see ads, Dexerto reports.

Trials of mRNA vaccines for lung cancer, pancreatic cancer and other diseases are already underway, The Washington Post reports.

The Information describes the early Anthropic investor’s bet on a moment he calls the “revenge of the scientists.”

The most advanced health-monitoring features require an extra subscription, TechCrunch reports.

The Wall Street Journal describes it as the first known breakout by Google’s AI. Google said it did not consider the episode an instance of model misalignment.