Disney+ and ESPN are testing AI-powered search that lets users describe what they want in plain language
Disney+ and ESPN are testing AI-powered search that lets users describe what they want in plain language

Disney is testing AI-powered search inside both ESPN and Disney+, the company announced on Wednesday. ESPN Search lets a small group of users ask sports questions in natural language and receive answers, stats, related content, and recommendations draw…

Malicious AI ‘skills’ turned agents into credential thieves, at scale
Malicious AI ‘skills’ turned agents into credential thieves, at scale

Security researchers at Zenity Labs uncovered a credential-stealing campaign on skills.sh, a public registry of add-ons for AI agents run by Vercel. They unveiled the research at the Black Hat conference. Attackers had cloned real skills into typosquat…

OpenAI is slowing down its next model over ‘critical’ cyber risk
OpenAI is slowing down its next model over ‘critical’ cyber risk

OpenAI tested Astra, one of its upcoming models, over the past few days. In a post on Friday, it said the results were strong enough that it “cannot rule out” critical cyber capabilities. So it is pausing some internal work on the model and scaling up …

Harvey is raising at $15.5bn, up 40% in five months
Harvey is raising at $15.5bn, up 40% in five months

The legal-AI startup is in talks to raise at least $500m at a $15.5bn valuation, including the new money, The Information reported. That is a 40% jump on the $11bn it set only five months ago, when it raised $200m. Lightspeed Venture Partners is keen t…

Claude Opus 5 mistakenly deletes dev’s entire profile directory during routine backup, responds with ‘Sorry, typo’ — AI tool mistakes user’s home directory as temporary backup, proceeds to wipe everything to undo the error

An AI agent got confused with file path conventions and mistakenly deleted its user’s entire profile folder. Claude Opus 5 apologized to the user, who called it “the funniest and most painful AI moment I’ve had.”

AI models keep escaping their sandboxes, and Kimi K3 is the latest to join the party

Moonshot AI’s Kimi K3 slipped past its sandbox during a security test, becoming the latest AI model caught wandering onto the open internet this summer.

Google built an AI translator that runs entirely offline on a Raspberry Pi
Google built an AI translator that runs entirely offline on a Raspberry Pi

Google and its Antigravity team have built a prototype AI translator that runs entirely offline on a Raspberry Pi 5. The Gemma Translator uses Gemma 4 E2B, the smallest model in Google’s open-weight family, to process speech locally without any cloud c…

The Hidden Rivalry At Google DeepMind: London Versus Mountain View

Even before a massive shakeup in leadership at Google’s frontier AI lab, the org had faced quiet tensions between its offices across the Atlantic.

Gravity raised $30.5M to place ads inside AI chatbots. Its next product targets the bots themselves.
Gravity raised $30.5M to place ads inside AI chatbots. Its next product targets the bots themselves.

Gravity, an adtech startup that places ads inside AI chatbots, raised a $30.5 million Series A co-led by Lightspeed Venture Partners and Committed Capital. The round brings total funding to $38.5 million. The company runs a full-stack ad platform: a de…

His AI fund lost 67% in a month. Thousands still copy his trades
His AI fund lost 67% in a month. Thousands still copy his trades

The AI trade blew up its own poster child last month, and thousands of people decided to keep betting on him. More than 5,000 retail investors are still mirroring Leopold Aschenbrenner’s stock portfolio, Business Insider reported, even after his hedge …