References

The Pentagon wants unrestricted access to AI technology. Anthropic is demanding guarantees against autonomous weapons control and domestic surveillance. A $200 million contract hangs in the balance.

Bytedance has released its new Seed2.0 AI model series. The story is a familiar one by now: the models match Western AI models on benchmarks while costing a fraction of the price.

Bytedance's Seedance 2.0 can generate Disney characters, replicate actors' voices, and recreate entire fictional worlds with stunning realism. Hollywood is fighting back with cease-and-desist letters and calls for legal actio, but the case highlights a growing problem: copyright law was built for a world where copying took effort.

The open-source framework Mastra compresses AI agent conversations into dense observations modeled after how humans remember things, prioritized with emojis. The system sets a new top score on the LongMemEval benchmark.

An AI agent wrote a hit piece on a developer who rejected its code. Days later, the agent is still running, a quarter of commenters believe it, and no one knows who's behind it. The case shows how autonomous agents turn character assassination into something that scales.

A new study reveals just how little it takes to shake up LLM rankings, raising fresh questions about how much weight the AI industry should put on (crowdsourced) benchmarks.