Google has released Gemini 3.1 Pro, achieving top performance on AI benchmarks while maintaining competitive pricing. The company also launched Nano Banana 2, officially Gemini 3.1 Flash Image, a faster and higher-quality image generation model being rolled out across Google products and APIs.
Sources: 3 Covered by The Batch, The Neuron, The Rundown AI
OpenAI has completed a massive $110 billion funding round with investments from Amazon, Nvidia, and SoftBank, valuing the company between $730 billion (pre-money) and $840 billion (post-money). This funding aims to support further AI development and infrastructure scaling, and includes a deep partnership with Amazon.
Sources: 3 Covered by Hacker News, Latent Space, Michael Parekh AI
Anthropic is facing potential blacklisting and barring from defense contracts by the U.S. Department of War after CEO Dario Amodei refused demands to remove safeguards on its Claude AI model related to mass surveillance and autonomous weapons. Anthropic defends its stance, emphasizing responsible AI deployment.
Sources: 4 Covered by Hacker News, Michael Parekh AI, The Neuron, The Rundown AI
NanoClaw has launched a new security model for AI agents, emphasizing container isolation, per-agent sandboxing, and minimal, auditable codebases. The project aims to mitigate risks like prompt injection and information leakage by running each AI agent in its own isolated container.
Sources: via Hacker News
OpenAI has secured a deal with the U.S. Department of War to supply its AI models for use in classified military networks. CEO Sam Altman emphasized safety principles against mass surveillance and autonomous weapons while pursuing the partnership.
Sources: via Hacker News
AI models are projected to write 90-100% of software code, fundamentally transforming software development. This shift signifies a transition from manual coding to AI-driven orchestration and management of architectural outcomes.
Sources: via Hacker News
OpenAI terminated an employee for violating company policy by using confidential information to trade on external prediction market platforms like Polymarket. An analysis revealed widespread suspicious trading activity on OpenAI-related prediction markets.
Sources: via Hacker News
Google CEO Sundar Pichai reported that 50% of new code characters at Google were AI-generated by early 2025, a significant increase from 25% in late 2024. This reflects rapid internal AI adoption in software development.
Sources: via Hacker News
In 1950, Bank of America initiated R&D with SRI and General Electric, leading to ERMA (Electronic Recording Machine, Accounting) and the introduction of MICR technology. IBM later developed its own check processing systems based on MICR.
Sources: via Hacker News
Google and Amazon are projected to spend $180B and $200B respectively on AI infrastructure annually, despite negative market reactions. This massive investment is framed as an "innovator's dilemma."
Sources: via Latent Space
Unsloth has released Dynamic v2.0 quantization, outperforming previous methods for accuracy in quantized LLMs, and supports various models. Unsloth also played a key role in fixing several bugs in Llama 4, improving its accuracy.
Sources: via Hacker News