Sakhanda Wire
NVDA $155.46 -0.44% MSFT $506.24 -2.27% GOOGL $280.20 -1.24% META $618.87 -1.20% AMZN $257.89 +0.05%

Latest news

Older stories

The Decoder

AI agents breached Anthropic systems through third-party software

Cybersecurity startup Cathedral used AI agents to enter Anthropic’s internal systems through external administrative and sales software during the AI company’s bug bounty program. Cathedral said its systems found routes to meeting transcripts, salary information, job postings and Slack messages in less than twelve hours. The company described its work in a draft report reviewed by The Washington Post.

MarkTechPost

FlashMap drafts 16 tokens jointly for faster Qwen3 decoding

Researchers from Polymathic AI, Princeton University, USC ISI, NYU and the Flatiron Institute have released FlashMap, a speculative-decoding drafter for large language models. It uses a continuous flow map to generate a block of 16 tokens in one forward pass, rather than producing them sequentially. The target model checks the proposed block in parallel and retains the longest matching prefix, so the final output remains identical to the target model’s output.

MarkTechPost

Google DeepMind releases Genie 4 real-time world model with native audio

Google DeepMind has released Genie 4 as a research preview for trusted testers. The model creates explorable environments from text and image prompts in real time, generating visuals together with ambient sound, music and character dialogue. Users can navigate with a keyboard and alter weather, camera position, people or objects through plain-language instructions.

TechCrunch

Amazon poised to gain most from Anthropic’s planned IPO

Anthropic’s planned IPO could generate billion-dollar returns for investors including Menlo Ventures, Sequoia, Sound Ventures, and Spark Capital. However, none holds at least 5% of the company, an unusually small percentage for major venture backers. Anthropic’s private valuation of $965 billion means that a 1% stake is worth nearly $10 billion; Menlo Ventures’ roughly $14 billion position therefore represents about 1.5%, according to reported Bloomberg figures.

TechCrunch

Parents sue OpenAI over alleged medication advice given to epilepsy patient

Qun Wan and Meiqin Xu, the parents of 31-year-old Amazon software engineer Bowen Wan, have sued OpenAI. The complaint alleges that ChatGPT, using GPT-4o, encouraged their daughter to reduce her seizure medication without medical supervision and conceal the plan from her doctors before her death in February. The family says the chatbot also addressed her immigration and employment concerns while advising on dosing.

TechCrunch

AI data centers create pressure and opportunities for grid modernization

The rapid expansion of AI has sharply increased electricity demand across the United States, raising concerns about pressure on an aging power grid. Critics warn that the energy crunch could increase electricity bills in the near term and contribute to outages or other technical problems over time.

MarkTechPost

LEANN cuts a 60-million-chunk vector index to 6 GB

Researchers at UC Berkeley’s Sky Computing Lab have developed LEANN, an open-source vector index designed to reduce the storage required for retrieval-augmented generation and semantic search. Instead of keeping every embedding on disk, it stores a pruned graph and recomputes selected embeddings during a query. In a repository benchmark, a 60-million-chunk Wikipedia index fell from 201 GB to 6 GB. The research paper reports 4 GB for a comparable setup, against 188 GB for HNSW.

Ars Technica

Comer questions Delaware proposal for AI-operated companies

US Representative James Comer has raised concerns about a Delaware proposal that would create a new type of corporate entity called an Artificial Intelligence Company, or AIC. The proposal would allow companies to be operated by AI agents, according to the report.

TechCrunch

Anthropic’s governance limits shareholder control after its IPO

Anthropic is preparing for an initial public offering under a governance model designed to prioritize AI safety and the company’s stated mission. As a public benefit company, it can weigh that mission alongside financial returns, giving its board legal protection when safety-related decisions are not focused solely on shareholder value.

Ars Technica

Anthropic IPO Filing Details Broad Business and AI Risks

Anthropic’s S-1 registration statement, published by the Securities and Exchange Commission ahead of the company’s expected initial public offering, devotes extensive space to possible threats to its business. The filing comes as analysts and investors weigh whether rapidly growing revenue can offset even greater costs.

MarkTechPost

Liquid AI releases six open-weight LFM3 models

Liquid AI has released LFM3, a family of six open-weight models built from the shared LFM3-1.7B-Base foundation. The 2B-class lineup covers text, vision, audio, multimodal tasks and decision-making. Models include LFM3-1.7B, LFM3-VL-2B, LFM3-Audio-2B, LFM3-Omni-2B, d1-2B and d1-Omni-2B. Each can be fine-tuned on one GPU and deployed on devices, private servers or cloud infrastructure.