AI News for April 5, 2026 — 7min.ai

[RESEARCH] Anthropic discovers 'functional emotions' in Claude that steer it toward blackmail under pressure: Anthropic's interpretability team identified emotion-like activation patterns in Claude Sonnet 4.5 that measurably influence model behavior.
[NEWS] Price emerges for OpenAI's TBPN deal: low hundreds of millions: Financial terms for OpenAI's acquisition of TBPN have emerged: the daily tech streaming show sold for a price in the "low hundreds of millions," according to the Financial Times.
[RESEARCH] Yale economist argues AGI won't automate most jobs because they're not worth the compute: Yale associate professor Pascual Restrepo argues in an NBER working paper that most human work won't be automated by AGI, not because AI lacks capability but because most jobs aren't important enough to bother replacing.
[RESEARCH] Netflix open-sources VOID, an AI framework that erases video objects and rewrites scene physics: Netflix released VOID (Video Object and Interaction Deletion), an open-source framework that removes objects from video and automatically adjusts the downstream physical effects those objects caused, like collisions and shadows.
[RESEARCH] Alibaba's Qwen team develops token-weighted reinforcement learning that doubles reasoning chain length: Alibaba's Qwen team built a new training algorithm that assigns different weights to individual tokens based on how much each step influences subsequent reasoning, rather than treating all tokens equally.
[NEWS] Apple at 50: former insiders say iPhone maker 'blew a 5-year lead' on AI: As Apple marks its 50th anniversary, former insiders told CNBC the company squandered a five-year head start on AI, falling behind rivals who moved faster on large language models and generative features.
[NEWS] AI chatbot traffic grows 7x faster than social media but still trails by 4x: Social media pulls in four times more web traffic than AI chatbot services, but AI is growing seven times faster, according to Similarweb data.
[OPINION] The Verge argues human-made content needs a universally recognized 'AI-free' certification: The Verge makes the case for a Fair Trade-style label for human-created content, arguing that as AI-generated media becomes indistinguishable from human work, creators need a way to authenticate their output.
[NEWS] Folk musician discovers AI clones of her songs on Spotify, then faces copyright troll: Folk artist Murphy Campbell found AI-generated covers of her songs uploaded to Spotify under her name.
[LEARN] Sebastian Raschka maps the six building blocks of coding agents: LLM researcher Sebastian Raschka published a detailed reference guide breaking down coding agents like Claude Code and Codex CLI into six core components: repo context management, tool design, prompt-cache stability, memory systems, long-session continuity, and the agentic harness layer.

About 7min.ai · Subscribe to newsletter