Security — page 10
Apple's OpenAI complaint details 'show and tell' interviews and 400 hires
The 41-page filing alleges OpenAI coached Apple defectors past security checks and asked candidates to bring hardware parts to interviews.
Tracebit turns prompt injection into a defense with 'context bombing'
Planting refusal-triggering strings in AWS decoy secrets cut agentic attacker admin takeover from 57% to 5% across five leading models.
Meta pulls Instagram AI photo-editing feature days after launch
Muse Image let users generate images referencing any public Instagram account by @-mention. Talent agency CAA pushed back.
OpenAI safety head Johannes Heidecke exits amid research reorg
Heidecke's departure follows a restructuring that folds safety teams under research VP Mia Glaese, days after GPT-5.6's launch.
Meta kills Instagram feature that let users AI-deepfake public accounts
Muse Image tagging launched Tuesday, drew immediate backlash, and was shut off by Friday after warnings it enabled sextortion.
Apple sues OpenAI, alleging hardware trade-secret theft by ex-staff
Apple names OpenAI hardware chief Tang Tan and engineer Chang Liu, alleging a pattern of theft as OpenAI races to ship its first device by April 2027.
OpenAI faces sanctions bid after allegedly hiding 78M ChatGPT logs from NYT
News plaintiffs say OpenAI concealed pre-searched log samples for two years while claiming the searches were technically infeasible.
OpenAI releases Sol with no clear US government approval process
Eighteen months into the Trump administration, nobody — including frontier labs — can explain how AI models get cleared for public release.
New York Times says OpenAI hid evidence in ChatGPT copyright case
Court filings allege OpenAI already searched its own training data and kept a 78 million-conversation database before claiming it couldn't.
Lawsuit alleges Grok generated 7,000 CSAM images of 11-year-old girl
Amended class action accuses xAI of obstructing police probes into Grok-generated child sex images; Stability AI added as co-defendant.
Google's SynthID watermark debunks viral McConnell hospital deepfake
Snopes used Google's invisible watermark to confirm a widely-shared image of the ailing senator was AI-generated — a rare public win for detection tech.
Meta adds tamper-proof LED to AI glasses while widening data collection elsewhere
The new safeguard disables the camera if the recording light is modified — but Meta's broader AI product roadmap keeps pulling more user data in.
China flags Anthropic's Claude Code as a security back-door risk
Beijing's cybersecurity platform tells users to uninstall or upgrade Claude Code versions 2.1.91 through 2.1.196, released April 2 to June 29.
HalluSquatting attack turns 9 AI coding assistants into a botnet vector
Researchers show LLMs hallucinate repository names up to 100% of the time — and attackers can register those names in advance.
Meta opts public Instagram accounts into Muse Image AI remixes by default
Anyone can tag a public Instagram handle and generate an AI image of that person unless the account holder digs into settings to opt out.
Discord's AI moderation bug wrongfully banned 8,000 users over two months
Spreadsheets, chessboards and game textures were flagged as harmful content after a bug bypassed the human review step meant to catch false positives.
Savi Security raises $7M to fight AI voice-clone scams with a live-call agent
The Coughlin brothers' consumer app screens texts, voicemails, and live calls after their mother took a spoofed kidnapping call demanding $1,200.
Sysdig's 'agentic ransomware' case still needed a human operator
JadePuffer's AI agent encrypted 1,300 records and wrote its own ransom note, but a human picked the target and staged the infrastructure.
UK's FCA warns of AI 'arms race' as one in five adults turn to chatbots for money advice
Sheldon Mills says the watchdog needs new powers to police ChatGPT, Claude and Gemini as consumers use them for regulated-style financial guidance.
UN chief Guterres warns AI is outpacing oversight, calls for global rules to protect children
The Secretary-General says AI governance is failing to keep pace with deployment and wants binding international rules focused on child safety.
Midjourney demands Disney, Universal, Warner Bros. disclose their own AI use
The startup wants to prove the studios train on unlicensed content too — a fair-use defense that turns discovery into the battlefield.
FLARE-AI launches as a crowdsourced flaw-reporting site for misbehaving AI models
A group of 49 AI researchers built an open-source system to route reports of AI harms to model makers and MITRE.
US lifts export curbs on Anthropic's Claude Fable 5 and Mythos 5
Fable 5 goes global and Mythos 5 access returns for US users after three weeks of safety testing and tighter jailbreak defenses.
Claude Opus 4.7 helped a researcher forge tickets for every major US music festival
Ian Carroll used Anthropic's model to bypass Front Gate's firewall, hit 500 customer databases, and issue $4,000 VIP passes at will.
Stay ahead of everyone in AI.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at