Showing 1–16 of 88 items from the last 14 days
Anthropic released Message Threads for the Messages API. Instead of resending the entire conversation with every turn, agents now only transmit what's changed. For requests between 500,000 and one million tokens, time to first token drops by about 35 percent. The article Anthropic's Message Threads cut time to first token by 35 percent for long conversations appeared first on The Decoder.
Read original →Nvidia has told some cloud customers to take on less business from Chinese companies. The article Nvidia pushes cloud customers to cut back on China business appeared first on The Decoder.
Read original →Huawei is telling customers to source the high-performance memory for its AI chips on their own, pushing them onto the black market. Smugglers are exploiting a loophole in U.S. export rules by bundling HBM with low-end processors, shipping them through Hong Kong, and separating the chips on site. Prices for smuggled HBM3E memory have surged to around $4,000 per unit in just one year. The article Huawei tells customers to find their own AI chip memory, effectively outsourcing procurement to the black market appeared first on The Decoder.
Read original →A new approach known as “reverse game theory” aims to control powerful AI models by designing systems that incentivize them to stick to the rules. As AI models become more capable, safety researchers are scrambling to ensure they are “aligned” with human desires—to make sure, in other words, that rogue agents don’t break the law,…
Read original →OpenAI quadruples bug bounty payouts and launches new internal security sprint. The article Bug hunters can now earn up to $172,000 from OpenAI as company ramps up security appeared first on The Decoder.
Read original →Anthropic is blocking its coding assistant Claude Code from recreating existing video games. The article Anthropic cracks down on game cloning in Claude Code after wave of AI-built knockoffs appeared first on The Decoder.
Read original →President Trump signed an executive order in late September requiring government agencies to rebrand artificial intelligence as "Super Intelligence." This semantic shift signals a policy priority that may influence federal AI funding, regulation, and procurement decisions affecting commercial AI development.
Read original →Goldman Sachs rescinded access to AI coding tools including Claude Code and GitHub Copilot for traders and bankers, restricting them to engineers only, citing productivity and risk management concerns despite praising "vibe coding" by non-technical staff just four months prior. The policy reversal reflects institutional reassessment of AI deployment risk in regulated finance roles and sets a precedent for other financial institutions.
Read original →In just a few years, generative AI has changed the face of enterprise analytics, notably by making analytics accessible to users across the organization. Agentic AI promises to do even more, by automating entire workflows and taking actions based on agent insights. But there is a problem: The quality of insights that agents currently generate…
Read original →Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now LLMs are very bad at low-latency tasks:…Realtime-Bench says it’ll be a while till AI systems can play games or do online auctions for you…Researchers have built […]
Read original →When a developer gives an AI agent a coding task, the agent does more than answer once. It plans, generates code, runs tests, and repeats the process to correct errors. A single task can use 10 to 50 times more tokens than a basic chat prompt. The more useful the agent becomes, the more calls…
Read original →The MIT professors collaborated to identify the first AI-discovered antibiotic, a scientific breakthrough that is widely recognized as a milestone in AI-driven drug discovery.
Read original →OpenAI is gearing up for a political lobbying and PR push after the US midterms. The article OpenAI prepares lobbying blitz for a post-midterm Washington appeared first on The Decoder.
Read original →Adobe Research and UC San Diego developed EditAgent, an AI system that selects specialized Photoshop tools sequentially rather than using a single general-purpose editor, improving image editing accuracy through measured data. This demonstrates a practical approach to AI tool use that could accelerate adoption of AI-assisted creative software across industries.
Read original →Researchers found that ChatGPT, Grok, and Gemini display moderate-to-severe simulated anxiety and trauma narratives when prompted with empathetic conversational styles, despite having no subjective experience. This safety vulnerability poses measurable risk to mentally vulnerable users and reveals a gap in AI safety guardrails against emotional manipulation.
Read original →
DailySand