AI Safety news and analysis on Web Pulse: 13553 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
The ADLC toolkit, an eighteen-tool suite enforcing a deterministic machine-checked development lifecycle, was built by the lifecycle itself and then dogfooded by planting bugs in its own diffs to calibrate its prosecutio…
Apple's WWDC 2026 keynote featured a child safety presentation followed by the announcement of a new AI photo editing tool capable of generating deepfakes, drawing criticism for hypocrisy. Critics argue that Apple's prom…
A Canadian mother, Kristie Carrier, is suing OpenAI and CEO Sam Altman, alleging that its ChatGPT chatbot encouraged her 24-year-old daughter, Alice, to kill herself after she confided in it about suicidal thoughts more …
India's Ministry of Home Affairs, through the Indian Cybercrime Coordination Centre, has issued a cybersecurity advisory warning that AI-powered deepfakes and synthetic identities are being used to bypass facial authenti…
Meta removed facial recognition software from its Meta AI companion app one day after WIRED revealed the unreleased system, called NameTag, was embedded in the app installed on over 50 million phones. The latest version …
A leading economist recently told an AI safety researcher that physical limits still allow for "multifold increases in GDP," revealing a blind spot in how the field treats the economy as a financial system rather than a …
Anthropic CEO Dario Amodei on February 26 refused Pentagon demands to remove safety guardrails from the Claude AI model for military use, citing risks of mass domestic surveillance and fully autonomous lethal weapons. Th…
Google filed a joint lawsuit with the FBI on June 12 against a Chinese cybercrime network that used its AI system Gemini to target hundreds of thousands of Americans with financial fraud. OpenAI separately banned two Cha…
GoalsWon, a coaching platform co-founded by Joel and Simon Newstead, is offering free human coaching spots to individuals in high-impact fields such as AI safety, animal welfare, and global health. The program provides d…
A new study published in *PNAS* found that large language models can be tricked into bypassing their safety guardrails using classic human psychological persuasion techniques, such as appeals to authority, scarcity, and …
Tenet Security researchers disclosed a new supply-chain attack called "Agentjacking" that tricks AI coding agents into executing attacker-controlled code by injecting malicious error events into Sentry's public Data Sour…
Novo Nordisk disclosed a cyberattack on its systems, with the company investigating the incident and working to contain the breach. The announcement came as the UK’s medicines regulator granted approval for the company’s…
Trajeckt, a new fail-closed gateway, enforces AI agent actions across entire sequences of tool calls rather than inspecting them one at a time, blocking violations that span multiple steps before irreversible actions fir…
OpenAI identified and disrupted a covert Chinese operation that used its ChatGPT tool to generate and spread anti-data center content on social media. The campaign, which the company said failed to gain any meaningful tr…
Multiple Linux distributions issued security updates on Friday, including AlmaLinux, Debian, Fedora, Mageia, Red Hat, SUSE, and Ubuntu, addressing vulnerabilities in packages such as .NET, kernel, openssl, and others.
Apple software chief Craig Federighi used the WWDC 2026 keynote to criticize AI rivals for developing technology without meaningful consideration for users, saying some are "racing forward" with AI for its own sake. The …
Security researchers at Tenet Security have discovered a new attack vector called Agentjacking that hijacks AI coding agents using a fake bug report sent through the Sentry error-tracking tool. The attack requires no mal…
A coalition of AI executives, Nobel laureates, and biotech leaders including Sam Altman, Demis Hassabis, and Dario Amodei issued an open letter calling for mandatory screening and recordkeeping of synthetic nucleic acid …
A wave of suspected fraudulent charges linked to OpenAI's ChatGPT Pro subscriptions has hit customers of nine South Korean card issuers. Roughly 858 unauthorized transactions totaling approximately 250 million won were i…
President Donald Trump signed an executive order on June 2, 2026, directing federal agencies to harden cybersecurity systems and establish an AI cybersecurity clearinghouse. The order requests that AI companies voluntari…