Musk testifies on Grok, Goodfire debuts Silico, Stripe tokenizes agent pay

Sources

Live updates from Elon Musk and Sam Altman’s court battle over the future of OpenAI theverge.com

Sam Altman and Elon Musk are facing off in a high-stakes trial that could alter the future of OpenAI and its most well-known product, ChatGPT. In 2024, Musk filed a lawsuit accusing OpenAI of abandoning its founding mission of developing AI to benefit humanity and shifting focus to boosting profits instead. The trial began with […]

Elon Musk testifies that xAI trained Grok on OpenAI models techcrunch.com

“Distillation” is a hot topic as frontier labs try to prevent smaller competitors from copying their models.

The craziest part of Musk v. Altman happened while the jury was out of the room theverge.com

Okay, I am not a lawyer so I only understood about half of what just happened. But I am fairly sure, given the context, that Elon Musk’s lawyers may have just fucked up big. Jared “James Brickhouse” Birchall, Musk’s finance guy and all-around fixer, took the stand after Musk today. Most of his testimony was […]

Elon Musk confirms xAI used OpenAI’s models to train Grok theverge.com

In a federal courtroom in California on Thursday, Elon Musk testified that his own AI startup, xAI, has used OpenAI’s models to improve its own. The matter at question is model distillation, a common industry practice by which one larger AI model acts as a “teacher” of sorts to pass on knowledge to a smaller […]

Elon Musk’s 7 biggest stumbles on the stand at OpenAI trial arstechnica.com

Elon Musk spent three days testifying as the first witness in his trial against OpenAI.

This startup’s new mechanistic interpretability tool lets you debug LLMs technologyreview.com

The San Francisco–based startup Goodfire just released a new tool, called Silico, that lets researchers and engineers peer inside an AI model and adjust its parameters—the settings that determine a model’s behavior—during training. This could give model makers more fine-grained control over how this technology is built than was once thought possible. Goodfire claims Silico…

Stripe updates Link, a digital wallet that autonomous AI agents can use, too techcrunch.com

Link lets users connect cards, banks, and subscriptions, then authorize AI agents to spend securely via approval flows.

Gemini is rolling out to cars with Google built-in theverge.com

Google is swapping the legacy Assistant for Gemini across millions of vehicles equipped with Google built-in, promising more natural conversations, vehicle-specific queries, and in-car settings control. The rollout marks the first time Gemini reaches automotive deployments at scale rather than via phone projection.

Google’s Gemini AI assistant is hitting the road in millions of vehicles techcrunch.com

The move signals Google’s push to bring more advanced, conversational AI into the driving experience.

Legal AI startup Legora hits $5.6B valuation and its battle with Harvey just got hotter techcrunch.com

Swedish legal AI startup Legora has raised at a $5.6 billion valuation, intensifying its rivalry with Harvey as both expand into each other’s home markets and trade dueling ad campaigns, including one starring Jude Law. Nvidia’s NVentures joined the round.

After dissing Anthropic for limiting Mythos, OpenAI restricts access to Cyber, too techcrunch.com

OpenAI will gate its new GPT-5.5 Cyber security-testing model to vetted defenders only at launch, mirroring the access restrictions it recently criticized Anthropic for placing on Mythos. The company had publicly called those limits anti-competitive weeks earlier.

Building gets easier bensbites.com

Ben Tossell walks through how his personal builder stack has shifted as agentic coding tools mature, arguing the bar to ship working software has dropped sharply and naming the specific tools replacing his previous workflow.

Meta is running get-rich-quick ads for its AI tools theverge.com

Manus, the AI agent startup Meta acquired for $2 billion last year, is paying creators to promote a pitch where users have AI build websites for local small businesses and cold-call the owners to sell them, drawing comparisons to get-rich-quick schemes.

(AINews) Agents for Everything Else: Codex for Knowledge Work, Claude for Creative Work latent.space

Latent Space’s daily AI digest argues coding agents are now “breaking containment” into general knowledge and creative work, contrasting OpenAI’s Codex push toward office tasks with Anthropic positioning Claude for writers, designers, and other non-developer users.

The hidden cost of Google’s AI defaults and the illusion of choice arstechnica.com

Ars Technica maps the opt-outs, dark patterns, and default settings that route user data into Gemini training across Google’s product suite, finding privacy controls fragmented across multiple dashboards and several toggles that re-enable themselves after feature updates.

Apple was surprised by AI-driven demand for Macs techcrunch.com

Apple said it will be supply-constrained on Mac mini, Studio, and Neo in the next quarter, too.

Sources: Anthropic potential $900B+ valuation round could happen within 2 weeks techcrunch.com

Anthropic is asking investors to submit allocations for the AI company’s latest fundraise within the next 48 hours, according to sources familiar with the matter.

X announces a rebuilt ad platform powered by AI techcrunch.com

X is rolling out a rebuilt ads platform powered by AI as it works to grow revenue again.

Meta says its business AI now facilitates 10 million conversations a week techcrunch.com

Meta said over 8 billion advertisers have used at least one of its GenAI tools

ChatGPT Images 2.0 is a hit in India, but not a big winner elsewhere, yet techcrunch.com

Users in India are embracing ChatGPT Images 2.0 for creative, personal visuals — from avatars to cinematic portraits.

(AINews) The Inference Inflection latent.space

a quiet day lets us reflect on the growing implications of the inference age

Meta cuts contractors who reported seeing Ray-Ban Meta users have sex arstechnica.com

Meta said the Kenyan workers didn’t “meet our standards.”

OpenAI announces new advanced security for ChatGPT accounts, including a partnership with Yubico techcrunch.com

OpenAI is launching additional opt-in protections for ChatGPT accounts. The new security initiative includes a new partnership with security key provider Yubico.

Salesforce is crowdsourcing its AI roadmap — with customers techcrunch.com

Salesforce lets its customers lead its product roadmap with the thinking that if one enterprise customer has a problem, the others likely do too.

FDA approval, fundraising, and the reality of building in healthcare according to BioticsAI founder techcrunch.com

BioticsAI CEO Robhy Bustami joined Isabelle Johannessen on Build Mode to discuss how the company has navigated a highly regulated space and kept the team motivated while cutting through all the red tape.

Builders bensbites.com

GPT-5.5 is a good model

(AINews) not much happened today latent.space

a quiet day.

References

The News (Pakistan) — recap of Birchall sidebar thenews.com.pk

I’m still struggling with how you can have conversations… but have no recollections even in a general sense

PYMNTS — judge on damages model pymnts.com

Judge Rogers has expressed public skepticism toward these figures, characterizing the $134 billion demand as ‘pulled out of thin air’

Berkeley Law — The Network blog on AI distillation sites.law.berkeley.edu

OpenAI relies on contract law rather than intellectual property (IP) law to prevent distillation… creates a ‘contractual moat’ where the developer owns the process and the platform access, even if they cannot legally own the specific text the model produces

Fast Company — model collapse risk fastcompany.com

training models on the outputs of other AIs can lead to ‘model collapse,’ a phenomenon where recursive training on synthetic data degrades a model’s reasoning and introduces ‘hallucinated’ artifacts

Kalshi prediction market news.kalshi.com

traders briefly priced Musk’s odds of a victory at roughly 53.5% following his initial testimony

Startup Fortune — coverage of admission startupfortune.com

Elon Musk admitted under oath that xAI may have trained on OpenAI’s models and then called it standard practice

TIME (Mayo Clinic / Arc Institute case study) time.com

Stanford’s James Zou cautioned that identifying a biological feature inside a model does not guarantee the model actually uses that concept to reach its final output.

Iván Arcuschin — MIB benchmark page iarcuschin.com

Sparse Autoencoders did not outperform standard neurons or raw hidden dimensions in the causal variable localization track

LessWrong / GreaterWrong — Anthropic JumpReLU discussion greaterwrong.com

Anthropic’s JumpReLU SAEs use learnable per-feature thresholds to avoid the shrinkage bias of L1 regularization, while Goodfire favors BatchTopK which relaxes the TopK constraint to the batch level.

Transluce — Predictive Concept Decoders transluce.org

PCDs translate a model’s internal states into human-readable concept lists to bypass deceptive self-reporting, positioned for scalable oversight of superhuman agents.

Longtermwiki E430 — Silico access notes longtermwiki.com

Silico is priced case-by-case; AutoSteer is throttled to ~30 requests per minute and a 50,000 token-per-minute global cap reflects the compute overhead of feature discovery.

GitHub gist (bigsnarfdude) — developer reaction notes gist.github.com

Developer feedback has been mixed, with some calling the platform ‘snake oil’ and others complaining of a clunky web interface filled with popups; over-steering can fundamentally break model reasoning.

Unite.AI — ‘Stripe hands AI agents a wallet’ unite.ai

Instead of the agent ‘seeing’ card details, it requests a spend authorization for a specific context; once approved by the user, the wallet issues a one-time-use virtual card or a Shared Payment Token (SPT) to complete the transaction.

GeekWire — ‘Judge blocks Perplexity’s AI bot from shopping on Amazon’ geekwire.com

A federal judge ruled that while the agent had the user’s permission, it lacked Amazon’s ‘authorization’ to access password-protected systems… Amazon argued that such agents ‘degrade the customer experience’ by bypassing personalization and fraud-detection systems.

PYMNTS — Mastercard Agent Pay pymnts.com

Mastercard has deployed Agent Pay, which utilizes ‘Agentic Tokens’ — dynamic digital credentials that extend tokenization to AI identities.

Corgi Labs — ‘Agent Payments Intelligence’ corgilabs.ai

When an agent initiates a purchase via Stripe’s Shared Payment Tokens, merchants lose access to critical telemetry like browser fingerprints, mouse movements, and typing cadences… agent-initiated charges are often structurally indistinguishable from standard card-on-file transactions in current reporting tools.

Taylor Wessing — ‘Agentic AI in Payments’ taylorwessing.com

Current consumer protections like Regulation E were designed for human-initiated transactions and lack a clear framework for ‘delegated’ AI spending… the EU’s revised Product Liability Directive currently excludes such monetary damages from its strict liability scope.

Payments Dive — Stripe/OpenAI ChatGPT checkout paymentsdive.com

High-profile merchants such as Etsy, URBN, Ashley Furniture, Coach, and Kate Spade have integrated Stripe’s Agentic Commerce Suite to make their catalogs ‘agent-discoverable’… Link currently serves over 250 million users.