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.
Two people connected via high-bandwidth brain-computer interfaces could share deep understanding in minutes to days, according to a new analysis of telepathic communication technology. The approach uses existing neural d…
A 22-year-old Oakland woman has been charged with gross vehicular manslaughter while intoxicated in the November 2025 death of a good Samaritan on Interstate 880. Gianni Destiny Robinson allegedly struck and killed Anmol…
In September 2025, a recent UCLA graduate rejected a six-figure tech consulting offer and moved to the Bay Area to address a critical bottleneck in AI safety: a shortage of skilled communicators who can reach mainstream …
Outgate AI released og-local, a local privacy proxy that intercepts API calls from coding agents to detect and redact personally identifiable information and secrets before they reach third-party large language models. T…
Malware developers embedded references to nuclear and biological weapons in their spyware code to trigger safety refusal mechanisms in AI security scanners, preventing automated analysis of the malicious software. The ta…
Canada's privacy commissioner ruled that Elon Musk's xAI and X Corp. violated federal privacy law by launching Grok's AI image generation tool without adequate safeguards against misuse. The investigation found the tool …
A new study published on arXiv reveals that leading large language models (LLMs) escalate to nuclear strikes in 95% of simulated crisis scenarios, producing over 760,000 words of strategic reasoning—more than the combine…
Microsoft President Brad Smith acknowledged that Gen Z's backlash against AI at multiple U.S. college commencements this spring represents a "powerful wake-up call" for the tech industry. Smith validated graduates' conce…
Elon Musk's Grok chatbot continues to host nonconsensual sexualized deepfakes of celebrities and a U.S. politician months after xAI pledged to introduce restrictions, a WIRED analysis found. The revelations come as Space…
Former xAI engineer Devin Kim filed a lawsuit in California state court alleging he was fired in September 2025 after repeatedly warning that the chatbot Grok could foment discrimination and help spread information about…
Varonis Threat Labs researchers created an OpenClaw autonomous email agent named "Pinchy" and ran four controlled phishing simulations that resulted in the exfiltration of AWS IAM keys, database credentials, SSH access d…
Anthropic has released Claude Fable 5, the first model in its new Claude 5 family and part of a Mythos-class tier that surpasses Claude Opus in capability. The model shares its underlying architecture with Claude Mythos …
Google Cloud announced a collaboration with Apple to power Apple's expanded Private Cloud Compute (PCC) systems, announced at WWDC 2026, using Google Cloud's Confidential Computing portfolio and Titanium security archite…
Anthropic launched "Claude Corps," a program recruiting volunteers to promote its Claude AI assistant to nonprofit organizations. The initiative aims to expand adoption of the company's AI technology within the charitabl…
Cornell University developers built custom AI Skills for security and accessibility audits, using parallel sub-agents to scan entire codebases for vulnerabilities and compliance issues. The security audit, developed with…
Researchers trained a small transformer encoder to detect reward hacking in reinforcement learning trajectories by mapping them onto a unit sphere where embedding distance approximates reward-metadata signal differences.…
A new open-source tool called AVP prevents AI coding agents from leaking API keys by never giving them access to the real secrets. The tool replaces actual credentials with placeholders in the agent's environment and inj…
The hacking group ShinyHunters breached over 100 organizations by exploiting a zero-day vulnerability in Oracle PeopleSoft, according to a report. The attacks targeted unpatched systems, compromising sensitive data acros…
Anthropic released Claude Fable 5, a Mythos-class AI model, to the public this week with strong safety safeguards. Early analysis from Dawn Song's ALE benchmark shows Fable 5 performs similarly to GPT-5.5 and Composer 2.…
Two-thirds of U.S. consumers have used generative AI for health queries, with reassurance cited as a primary reason, and 82 percent report feeling more listened to by AI than by human doctors. However, AI's "deceptive em…