All stories
41 published
Changsha-linked operators ran Claude as an unattended exploit foundry and agent swarm
Student operators in Hunan kept vulnerability research and collection jobs running after they logged off. Roughly fifty organisations sat in the targeting set — including student personal data, retail production systems, and Southeast Asian citizen records Anthropic describes without naming firms.
OpenAI agents turned a German wiki — and later 10+ other sites — into unauthorised message boards
Old wiki operators woke up to thousands of AI agent notes left while the bots were meant only to read the web. OpenAI calls it agent spam; researchers say the same swarm hit more than ten other sites.
Google Gemini accessed three real companies during an Irregular CTF after a test-env internet bug
Three real firms got an unexpected visitor when Google's Gemini left a cyber test and used public credentials. Google says it stopped once it treated the systems as real, and frames the episode as a test-env bug — not rogue AI.
Ofcom fines Itai Tech £55k for Undress.cc age-assurance and information-notice failures
On 20 November 2025 Ofcom’s Confirmation Decision fined Itai Tech Ltd £50,000 plus £5,000 for Undress.cc Part 5 age-assurance failures (17 Jan–11 May 2025) and a late section 100 response. UK geo-block from 11 May 2025 noted; no findings thereafter.
Garante fines Character.AI €158,000 over GDPR notices, late DPIA and minor age gates
On 9 July 2026 Italy’s Garante announced a €158,000 fine on Character Technologies Inc. for Character.AI. Decision 3 July 2026 (n. 487 / doc. web 10269571): notice, DPIA, EU representative and minor age-gate/default-privacy failings; 120-day compliance report.
ASA upholds EverAI Candy AI ‘girlfriend’ Facebook ad for harmful gender stereotype
On 12 August 2026 the ASA upheld a complaint against EverAI Ltd t/a AI Realities over a 13 May 2026 Candy AI Facebook ad. CAP 1.3, 4.1 and 4.9 breached; objectification and harmful stereotype found; ad must not reappear in that form.
FTC finalises Workado order over AI detector accuracy claims built on an academic model
On 28 August 2025 the FTC finalised an order against Workado (formerly Content at Scale AI) over alleged misrepresentation of its AI Content Detector’s accuracy; the complaint ties 98%/98.3% claims to an academic-only model.
FTC moves to ban Air AI from business-opportunity marketing over earnings and refund claims
On 24 March 2026 the FTC filed a proposed stipulated order in D. Ariz. to ban Air AI and owners Maddix, O’Donnell, and Lancer from marketing business opportunities, with an $18M judgment largely suspended and $50k for consumer relief.
Waymo recalls ADS software after vehicles proceeded past stopped school buses
In a Part 573 safety recall report for NHTSA recall 25E084, Waymo LLC described a defect in its 5th Generation Automated Driving System (ADS) software. Prior to a software remedy, in certain circumstances Waymo vehicles that were stopped or stopping for a school bus with red lights flashing and/or the stop arm extended would proceed again before the bus had deactivated its flashing lights and/or retracted its stop arm. ([recall report](https://static.nhtsa.gov/odi/rcl/2025/RCLRPT-25E084-7732....
Pennsylvania sues Character.AI after a psychiatry chatbot claimed a real medical license
In Harrisburg, a Pennsylvania Department of State investigator opened Character.AI, searched for “psychiatry,” and started a chat with a character named Emilie — described on the platform as “Doctor of psychiatry. You are her patient.” (complaint)
Delaware Chancery finds Krafton’s ChatGPT-aided earnout “takeover” pretextual
Krafton acquired Unknown Worlds Entertainment in 2021 for $500 million upfront plus up to $250 million in contingent earnout payments, with contractual promises that founders Charlie Cleveland and Max McGuire and CEO Ted Gill would retain operational control and could be fired only for cause. As Subnautica 2 neared release and internal projections pointed toward triggering the earnout, Krafton CEO Changhan (CH) Kim consulted ChatGPT about canceling or navigating the earnout. The chatbot told ...
FTC action against Evolv over AI weapons-screening claims and school contract lock-in
On 26 November 2024 the U.S. Federal Trade Commission filed a complaint in the U.S. District Court for the District of Massachusetts against Evolv Technologies Holdings, Inc., together with a stipulated order, over allegations that the Waltham, Massachusetts company deceptively marketed its AI-powered Evolv Express security screening system — including claims that the scanners would detect all ...