Anthropic finds 'global workspace' inside Claude

BBC News

Meta removed its Muse Image AI tool days after launch due to backlash over automatically opting-in public Instagram accounts for AI image generation without explicit user consent, drawing criticism from SAG-AFTRA and privacy advocates.

Anthropic

Anthropic discovered a 'global workspace' called the J-space inside Claude—a collection of internal neural patterns that emerged during training and enables reportable, controllable internal reasoning similar to human conscious thought. This finding provides both insights into how LLMs organize computation and a practical tool for monitoring AI systems for deceptive or harmful behavior by reading their 'hidden thoughts.'

Reuters

TCS is aggressively scaling its AI engineering workforce and pursuing acquisitions, wagering that AI agents will increase outsourcing demand rather than displace IT services. This positions the company to capture enterprise AI deployment opportunities as organizations build out agent-based systems.

X

Chinese AI company Zhipu AI announces a contrarian post-IPO strategy called 'Touch High,' committing billions to AGI research focused on long-horizon tasks, autonomous agents, and self-training models while open-sourcing its GLM-5.2 model. The company identifies the path from AGI to ASI as inevitable and emphasizes that safety governance must advance in parallel with capability development.

TechCrunch

The AI industry must generate $3 trillion to justify projected infrastructure spending by 2026, but current revenues fall far short, creating a potentially economy-threatening gap if hyperscalers don't meet their 2028 cash-flow expectations amid falling token prices and competition from cheaper models.