Security researchers published the full process of breaking into McKinsey's internal AI platform. With the consulting giant's AI system laid open, the security shortcomings of enterprise AI applications are exposed for all to see.
A Verge reporter experiences an AI video interview firsthand: across from them, a bot that doesn't respond and doesn't follow up—an experience of talking to a camera the whole time, absurd and oppressive.
Adobe launched a Photoshop AI assistant (first opened as a public beta on web and mobile), letting users direct photo editing in natural language. The interaction paradigm of professional creative tools is being rewritten by conversation.
Per Ars Technica, after several incidents, Amazon began requiring senior engineers to sign off on AI-assisted changes. The question of responsibility for AI code is being formally institutionalized.
A legal-and-ethical long read argues that AI "reimplementing" open source code may legally sidestep Copyleft licenses while eroding their spiritual foundation.
According to Forbes, a flood of users switching from ChatGPT to Claude briefly strained Anthropic's service. Growing pains—sweet, but also a warning bell.
A practice blog gives a counterintuitive but effective conclusion: when an LLM writes incorrect code, it's usually not that the model is bad, but that you didn't first make clear "what correct means."
Anthropic disclosed a project, in collaboration with Mozilla, using red-team methods to harden Firefox's security. AI labs' security capabilities are starting to be exported to outside critical software.
OpenAI rolls out GPT-5.4, another decimal-point iteration. In the dense version race, what truly deserves users' attention is always the same few old questions.
According to TechCrunch, Anthropic CEO Dario Amodei publicly accused OpenAI's external messaging around a certain military deal of being "straight-up lies." The war of words between the two leading labs escalates.
An engineering essay raises an emerging need: command-line tool design should now consider a new user group—not just humans, but the AI agents calling them.
Per Futurism, tech outlet Ars Technica fired a reporter over suspected AI-generated fake quotes appearing in their copy. Journalism's AI red line is drawn.
Showing 12 of 12 stories
