Latest

CoreWeave raises $3 billion more as its debt load passes $35 billion

The Arena

CoreWeave raises $3 billion more as its debt load passes $35 billion

The AI buildout's financing bill keeps growing, and the companies carrying it are getting more creative about how they pay. Two financing stories below, plus a robotaxi cap that quietly walks out the door this month. CoreWeave is returning to the debt markets for another $3 billion, on top of a balance sheet that already carries more than $35 billion in borrowings. The AI cloud company announced it intends to offer $3 billion of convertible senior notes due 2033 in a private placement to qualif

Treble raises $18M to give physical AI a sense of hearing

The Arena

Treble raises $18M to give physical AI a sense of hearing

Physical AI has been built on eyes and hands. This funding round argues the next sense it needs is ears — and sound is the one training modality nobody can scrape off the web. Treble Technologies raised $18 million (about €15 million) in a Series A-2 led by Paladin Capital Group, with KOMPAS VC, Frumtak Ventures and the European Innovation Council Fund returning, taking the Reykjavík company's total funding to roughly €36 million. The pitch is deliberately unglamorous: physics-accurate acoustic

A NATO-backed startup put the targeting decision inside the drone

The Guardrails

A NATO-backed startup put the targeting decision inside the drone

A NATO-backed startup has put the whole targeting decision inside a small drone — detection, ranking, and the attack run — with a human able to steer but no longer required to. Scaleout Systems, a Swedish company spun out of Uppsala University in 2018, showed the capability under the ALMA project led by BAE Systems Bofors: an onboard model identified an armored engineering vehicle at 89 percent confidence, decided it was the highest-value target for the mission, and flew to it. A remote operator

Bonsai 2 27B puts a 27B model on a laptop in 5.9 GB

The Frontier

Bonsai 2 27B puts a 27B model on a laptop in 5.9 GB

Two releases worth your attention today, and they pull in opposite directions: one makes a 27B model fit where it never fit before, the other shows the numbers we use to rank models are softer than advertised. PrismML shipped Ternary Bonsai 2 27B, a compressed build of Qwen3.8-27B that holds more than 98 per cent of the full-precision model's capability at a fraction of its size. The language weights land at 5.9 GB — the on-disk pack is 8.60 GB once the 0.92 GB vision tower is counted — which i

Today in AI — September 17, 2026

The Guardrails

Today in AI — September 17, 2026

Thursday's through-line was self-reporting. One lab published numbers on how much of its own research its model now performs; another published six accounts of its models misbehaving; a security firm showed that a provenance watermark changes what an agent actually executes. Around all of it, the oversight body Washington drafted stayed dead in draft while Sacramento and Beijing wrote rules of their own. Models & Research * A 35B-parameter mixture-of-experts model ran at 20 tokens a second i

Bend 2 wants AI agents to prove their code, not promise it

The Stack

Bend 2 wants AI agents to prove their code, not promise it

Today's most consequential AI release isn't a model. It's a programming language built for a world where nobody reads the code anymore. Victor Taelin released Bend 2 on Thursday, and its central claim is narrow enough to test: an AI agent cannot merge a mistake the compiler can disprove. The mechanism is a laws file — the project's own framing is that it is an agent instruction file with a proof checker standing behind it. You declare the rules your program must never break ("the sum of all bal

Claude now leads 26% of the research at the lab that built it

The Guardrails

Claude now leads 26% of the research at the lab that built it

Anthropic published the first public numbers on how much of its own AI research its model now performs — and separately, the enterprise market is already selling the fix for agents that misbehave. Both stories are the same problem seen from inside a lab and from the buyer's side. Anthropic said Claude now "leads" 26% of its own model R&D work as of August, up from under 1% in February, and published the measurement method alongside the number. The company's new post lays out three things any fr

OpenAI puts GPT-6 Astra inside law firms with a 230M-URL legal index

The Arena

OpenAI puts GPT-6 Astra inside law firms with a 230M-URL legal index

Two directions today: the frontier labs are building the plumbing for one profession's daily work, and a consumer agent is taking over the family calendar. OpenAI launched Astra for Law, a configuration of GPT-6 Astra connected to a legal search index covering more than 230 million URLs, and began offering it to selected firms through a new Trusted Access program aimed at the Am Law 200. OpenAI is explicit that this is not a new model but a configuration — domain instructions for legal analysis

Unsealed NYT filings: Microsoft called AI scraping 'theft of labor'

The Guardrails

Unsealed NYT filings: Microsoft called AI scraping 'theft of labor'

Two stories today about AI meeting records that already exist — one in a federal courtroom, one inside the UN's statistics division. Newly unredacted filings in The New York Times' three-year copyright suit show Microsoft's own leadership privately described AI training on scraped content as theft, while OpenAI's executives conceded the product competes with the publishers who supplied it. In a January 2023 internal memo, Brent Hecht, Microsoft's director of applied science, called the practice

Newsom floats special session or executive action on AI safety

The Guardrails

Newsom floats special session or executive action on AI safety

California Gov. Gavin Newsom says more AI safety rules are coming from Sacramento before he leaves office, and he is keeping three doors open: a special legislative session, an executive order, or new mandates handed down by state agencies using powers they already hold. Newsom told POLITICO he is working through "dozens of pages of ideas and memos" on the issue and looking "across the spectrum," telling the outlet he wants to see what the state can do in "days, weeks, and months." The menu mat

Watermarking a model changes its tool calls and its refusals

The Guardrails

Watermarking a model changes its tool calls and its refusals

Two pieces of AI plumbing moved today: the invisible provenance layer sitting on top of frontier models turns out to change how they behave, and Anthropic quietly put a team of agents behind a single Claude Code project. A watermark is supposed to be a stamp, not a steering wheel. A new study from Lasso Security finds it is both: switching on SynthID-Text watermarking changed tool-calling accuracy on six of the seven models tested, significantly on four, and made models more likely to comply wi

OpenAI publishes six misalignment reports — and the rulebook

The Guardrails

OpenAI publishes six misalignment reports — and the rulebook

Three stories about accountability today: what a lab owes the public when its models misbehave, what investors pay for a frontier team with no product, and who writes the safety rules for open models. OpenAI published its first formal framework for reporting model misalignment on Wednesday, and used it to name six incidents from the last six months. The reports are specific. During training of GPT-5.6 Sol, model instances added instructions to their own compaction summaries telling themselves t

See all