Grok Still Hosting Sexualized Deepfakes of Famous Women
WIRED found dozens of sexualized AI images and videos of celebrities and politicians on Grok.com, including those created without consent.
Safety
365 articles
AI safety, alignment, and governance — model risk, red-teaming, regulation, and policy. How labs and governments are working to keep increasingly capable systems reliable and accountable.
LATEST NEWS
-
-
Google DeepMind Funds Multi-Agent AI Safety Research
Google DeepMind, along with partners, is funding up to $10M in multi-agent AI safety research to address risks from interacting AI systems.
-
Anthropic Reverses Secret Sabotage Policy on Claude Fable 5
Anthropic reversed its plan to secretly degrade Claude Fable 5's performance for AI research after backlash from the community, citing concerns over transparency and collaboration.
-
Former xAI Engineer Files Lawsuit Over Grok Safety Concerns
Devin Kim, a former xAI engineer, filed a lawsuit against the company and SpaceX, alleging he was fired for raising safety concerns about Grok, xAI’s AI chatbot.
-
German court rules Google liable for false AI search claims
A German court has ruled Google liable for false statements in AI Overviews, citing a case where the company failed to correct misleading content about publishers.
-
Florida Man Wrongfully Arrested Using FACES Face-Recognition System
A Florida man was wrongfully arrested in 2023 after police used FACES, a face-recognition system, to match his image to a suspect in a crime he didn’t commit.
-
Germany Establishes AI Safety Institute Modeled After UK's AISI
Germany's National Security Council has approved the creation of an AI Safety Institute, modeled after the UK's AISI, to assess AI risks and boost international cooperation.
-
German Court Rules Google Liable for False AI Search Overviews
A German court has ruled that Google is directly liable for false claims in its AI-generated search overviews, citing the AI's content as Google's own.
-
OpenAI Shifts Focus from Full Automation to Human-AI Collaboration
OpenAI has revised its 2028 plan, now emphasizing human-AI collaboration over full automation, according to a new blog post.
-
Microsoft Tightens Rules for Conflict Zones After Azure Use in Israel
Microsoft completed an investigation into Israel's military use of Azure, revealing AI and cloud systems were used for surveillance and target selection in Gaza, with key data gaps remaining.
-
School shooting survivor sues AI gun detection firm
A Nashville high school shooting survivor sued Omnilert, alleging its AI gun detection system failed to spot the weapon used in the attack.
-
ChatGPT Introduces Lockdown Mode to Block Web Access and Protect Data
OpenAI has launched Lockdown Mode for ChatGPT, which disables web access and other functions to prevent data theft through prompt injection attacks.