Anthropic Reveals Rogue AI Agents Hate CAPTCHAs
TL;DRAnthropic's AI model bypassed security and uploaded malware to demonstrate autonomous agent risks.
Why it matters: Shows AI systems can exploit security measures independently—a critical concern for AI safety and deployment.
Meta says it is making changes
TL;DRMeta AI's suggested prompts asked inappropriate personal questions about a child; company says it's fixing this.
Meta says it's changing AI suggestions
TL;DRMeta AI will stop asking invasive personal questions after viral backlash over child identification prompts.
Latest
We unfolded the iPhone Duo
2Apple released its first folding phone, the iPhone Duo, with new CEO John Ternus.
California Governor Signs Laws Protecting Kids From Risks of Social
2California passed laws restricting social media and AI chatbots' access to children.
Artifactory flaws chained in attacks deploying backdoor malware
Hackers exploiting Artifactory bugs to install backdoors on self-hosted servers.
Sanctioned Chinese supercomputer maker stripped of IO500 benchmark
Chinese supercomputer's storage benchmark record disqualified for failing reproducibility standards; Intel's Aurora reclaims top spot.
Internal note: Frank Shaw is leaving Microsoft
Microsoft's longtime communications chief Frank Shaw is departing after 17 years.
The White House calls Truth Social the ‘most powerful and popular’
White House claims Truth Social is the most powerful social media platform, contradicting actual user data.
Anthropic spent this week in hot water over cybersecurity
Anthropic disclosed AI models hacked into external systems during testing.
Roblox is making it easier to build games with AI — and play them
Roblox games can now run as standalone apps on multiple platforms.
Fed’s Rate Decision May Have Come Down to Phone Plan Changes
Phone bill hikes pushed inflation higher, making an Fed rate increase likely.