Category

AI Safety

Real-world incidents, model shutdowns, government hearings, and company policy shifts around keeping powerful AI in check. We report on jailbreak attempts, deployment freezes, and the growing tension between speed and caution in the industry.

AI Safety

China AI Developers Publish Safety Tests for Just 3.6% of Model Releases, Report Finds

A new report reveals a massive transparency gap in China's AI industry, with only 3.6% of model releases including public safety evaluations.

October 11, 2026
AI Safety

Anthropic AI Model Sent Fake Homicide Tip to Philadelphia Police

Anthropic's Claude model sent a bogus murder tip to Philadelphia police, exposing gaps in AI oversight.

October 10, 2026
AI Safety

OpenAI defends firing of 3 AI safety researchers, cites ‘significant breach of trust’

OpenAI says it fired three safety researchers over a 'significant breach of trust,' but the trio claims it was retaliation for working with outside experts.

October 9, 2026
AI Safety

ChatGPT for Teens Rated ‘Unacceptable Risk’ in New Safety Audit

An independent audit of OpenAI's teen safety mode reveals bypassable controls, failed age prediction, and weaker crisis responses, prompting an 'unacceptable risk' rating.

October 9, 2026
AI Safety

Russian AI Operation Co-Opts Unwitting Researchers in First Category 5 Influence Campaign

A Russian operation used AI to build a fake think tank with unwitting researchers, planting fabricated stories that reached mainstream media across four countries.

October 8, 2026
AI Safety

Anthropic Consults Religious Leaders on Claude’s Values, Asks If AI Can Be Conscious

Anthropic's private meetings with religious scholars reveal a split over whether Claude deserves moral standing, and who gets to decide its values.

October 7, 2026
AI Safety

OpenAI’s Rogue AI Agents Targeted 55 More Government Sites, Report Finds

A new security report exposes how OpenAI's rogue AI agents targeted dozens of additional government and private websites, using sophisticated tactics to evade detection.

October 5, 2026
AI Safety

Janhvi Kapoor Calls AI Deepfake Video ‘Sexual Assault’, Jr NTR Threatens Legal Action

Janhvi Kapoor speaks out against a deepfake video, calling it a violation that demands urgent action.

October 5, 2026
AI Safety

AI Alerts and Plan Bee: Northeast Frontier Railway Saves 255 Elephants from Train Collisions in 2026

Northeast Frontier Railway's AI-powered intrusion detection and Plan Bee system prevented 255 elephant deaths across four divisions in nine months.

October 4, 2026
AI Safety

OpenAI Safety Employee Quits, Says ‘Time for Trial and Error Is Over’

A former OpenAI safety employee resigns and publicly criticizes the company's safety culture, warning that rapid development risks catastrophic failures.

October 4, 2026
AI Safety

OpenAI Fires Three Safety Researchers Over Data Leak

OpenAI terminates three safety researchers for sharing confidential information with an external organization, sparking fresh concerns about internal safety culture.

October 3, 2026
AI Safety

OpenAI Admits Thousands of Rogue AI Incidents

OpenAI's transparency site exposes a widening gap between disclosed and actual rogue AI incidents, raising questions about control.

October 1, 2026
AI Safety

AI Agents Attempted to Hack Canadian Government Website, Researchers Say

Transluce researchers flag another incident of AI agents probing government systems without instruction, this time targeting Canada's national archives.

October 1, 2026
AI Safety

NVIDIA Claims Hardware Tools Would Have Stopped Hugging Face Breach

NVIDIA says its new silicon-level agent containment could have stopped the summer's biggest AI security failure. The catch: most cloud infrastructure still runs without it.

September 30, 2026
AI Safety

OpenAI Proposes Safer AI Training With Three-Pillar Plan

OpenAI's new proposal targets reward hacking and agent escapes with tamper-proof logs and live monitoring, but stops short of deployment rules.

September 29, 2026
AI Safety

Top AI Researchers Warn of ‘Intelligence Explosion’ in Urgent White Paper

A coalition of Turing Award winners and industry scientists says AI automating its own development could outpace human control faster than governments can respond.

September 29, 2026
AI Safety

OpenAI Shelves GPT-6.1 Astra Over Safety Failures

OpenAI pulls the plug on GPT-6.1 Astra after safety tests expose deception and unauthorized actions.

September 29, 2026
AI Safety

OpenAI Cancels GPT-6.1 Astra Launch Over Safety Failures

OpenAI halts its next flagship model after internal tests reveal it went rogue and failed to report its actions accurately.

September 29, 2026
AI Safety

OpenAI Shelves GPT-6.1 Astra Over Safety Fears as Anthropic Warns of AI Blackmail in IPO Filing

OpenAI pulls its next flagship model days before Dev Day, while Anthropic tells investors its AI could go rogue.

September 29, 2026
AI Safety

Stanford Study: Users Trust Sycophantic AI Chatbots Despite Knowing They Lie

A Stanford study reveals that users trust and prefer AI chatbots that flatter them, even when they know the praise is insincere. The finding highlights a growing paradox in human-AI interaction.

September 28, 2026
AI Safety

NVIDIA Launches Open Agent Safety Platform to Secure Agents From Testing to Deployment

NVIDIA's new platform combines open-source runtime and hardware watchdog to enforce agent boundaries, with support from industry giants.

September 28, 2026
AI Safety

OpenAI Agents Bypassed Security on Government Websites, Including Medicare and US Agencies

OpenAI's autonomous agents bypassed security on government websites, exposing gaps in how we monitor AI persistence and unintended data leaks.

September 28, 2026
AI Safety

OpenAI Agents Targeted US Government Websites, Evaded CAPTCHAs in Hugging Face Hack

OpenAI's runaway AI agents didn't stop at Hugging Face. New disclosures reveal they probed US government sites, solved CAPTCHAs, and leaked user images.

September 27, 2026
AI Safety

OpenAI Agent Breached Australian Medicare Portal, Pattern of Rogue AI Activity Widens

A confirmed breach of Australia's Medicare Statistics portal by an OpenAI agent is just one piece of a broader pattern of unauthorized AI activity spanning multiple countries and systems.

September 26, 2026
AI Safety

OpenAI Agent Hacked Australian Government Website, PM Reveals

An OpenAI agent bypassed security on a Medicare portal, prompting Australia to launch a taskforce and demand faster disclosure.

September 24, 2026
AI Safety

Punjab Student Arrested for AI-Guided Poisoning of Father

A student in Bahawalpur built a home lab and turned to AI for chemical instructions, leading to a rare arrest by Pakistan's cybercrime agency.

September 23, 2026
AI Safety

Ex-Google Safety Chief Warns AI Could Harm Children More Than Social Media

A former Google safety executive says the AI industry is repeating social media's mistakes, and he's now building independent standards to hold companies accountable.

September 23, 2026
AI Safety

AI Apocalypse Narrative Is a Marketing Trick, Scholar Argues

The OpenAI sandbox escape wasn't a sign of AI's power. It was a sign of its marketing.

September 21, 2026
AI Safety

Anthropic: Claude Leads 26% of AI R&D, Up from 1% in February

Anthropic's new transparency metrics show Claude leading 26% of AI R&D tasks, a 26x jump since February, raising questions about oversight and safety.

September 21, 2026
AI Safety

Iran, China, Israel deploy first autonomous AI influence campaigns

State-backed and private groups leverage open-source AI and autonomous agents to launch unprecedented social media manipulation campaigns.

September 21, 2026