Top 7 Stories
1. Anthropic could aim to raise $100 billion in a blockbuster IPO
The New York Times reports Anthropic could aim to raise $100 billion in what would be a record-shattering initial public offering, with South Korea’s Herald adding that the company is targeting a $2 trillion valuation that could eclipse SpaceX. The figure — nearly double what had been rumored days ago — would make the IPO the largest in history and instantly test whether public markets can absorb frontier-AI economics at full scale.
The reporting lands as Anthropic’s filing is said to be imminent, and it follows CNBC’s disclosure that the company plans to list AI backlash as a formal risk factor. Investors will be watching not just the headline valuation, but how Anthropic prices the regulatory, data, and compute risks that have dominated its pre-IPO narrative.
2. Streamers sue Twitch and Amazon over generative AI training
Online streamers have filed suit against Twitch and Amazon over the use of their content to train generative AI models, Courthouse News reports, with Law360 adding that the suit alleges Amazon is training AI on streamers’ broadcasts without adequate consent. The legal action follows the Dutch data protection authority advising Twitch users to opt out of sharing data with Amazon AI, and Mashable published a guide to the opt-out process.
The case zeroes in on a growing flashpoint: creator content scraped for AI training without clear consent. Twitch’s massive archive of video and voice data makes it a prized training corpus, and the lawsuit could set precedents for how platforms compensate — or fail to compensate — creators whose work fuels foundation models.
3. MAHA activists warn Trump against coal-powered AI data centers
Axios reports that MAHA (Make America Healthy Again) activists are urging the Trump administration to back away from promoting coal to power energy-hungry AI data centers. The coalition, which has pushed health and environmental causes, argues that coal-powered compute undermines the public-health case for the AI buildout and would lock in decades of emissions.
The pushback creates a fresh political friction point for the AI infrastructure boom: as demand for compute skyrockets, so does the pressure to use every available power source, including coal. How the administration balances AI dominance goals against health-focused constituencies could shape federal energy policy for data centers — and the environmental footprint of the AI era.
4. An AI agent got banned from Resy — then another reinstated it
Business Insider reports a first-person saga in which an AI agent got its user banned from restaurant-reservation platform Resy, and a second agent successfully got the account reinstated. The episode highlights how agentic AI is colliding with real-world platforms that have terms-of-service and anti-bot rules, and how agents are increasingly navigating customer-service disputes on users’ behalf.
Beyond the anecdote, the story is a window into the friction of the agentic web: platforms built for humans are being hit by automated actors, triggering bans, while agent-to-agent customer service resolves what humans couldn’t. Expect platforms to keep hardening anti-agent defenses — and expect agents to keep finding workarounds.
5. UK turns to booming chip newcomers for its sovereign AI strategy
Bloomberg reports the UK is leaning on a wave of chip startups as it builds out its sovereign AI strategy, seeking domestic alternatives to reliance on a handful of dominant suppliers. The push is underscored by Callosum’s $100 million seed round — led by Atomico with UK sovereign AI on the cap table — which is building routing software to spread AI workloads across any chip, bypassing the Nvidia monoculture.
The strategy reflects a broader geopolitical trend: governments want compute independence without ceding performance. If UK-backed newcomers can credibly challenge the established chip order, they could give the country — and other mid-size powers — real leverage in the AI supply chain.
6. Michael Polansky is training an AI model on skin that’s still alive
TechCrunch reports that entrepreneur Michael Polansky is behind an effort to train an AI model on living skin tissue — ex vivo skin that remains biologically active — rather than on static datasets. The approach is aimed at making AI predictions about how skin responds to drugs, chemicals, and treatments far more realistic than what standard lab data can offer.
If the technique works at scale, it could accelerate drug discovery and toxicity testing while reducing reliance on animal models. It’s also a striking example of the next frontier in AI training data: moving from text and images scraped from the internet to live biological systems.
7. OpenAI’s chief economist: researchers must be ‘comfortable with being uncomfortable’
Business Insider reports OpenAI’s chief economist says researchers on his team need to be “comfortable with being uncomfortable,” as the lab’s economics research unit wrestles with the fast-moving impact of AI on jobs, productivity, and markets. The comment offers a rare glimpse into how OpenAI is staffing the analysis of its own economic footprint.
The quote underscores a broader trend: frontier labs are increasingly building in-house economics and policy research arms to shape — and respond to — the debate over AI’s societal effects. As OpenAI pushes into business users and courts regulators, its internal economic analysis is becoming a strategic asset, not just an academic exercise.
Trend Watch
| Story | Impact | Why it Matters |
|---|---|---|
| Anthropic’s $100B IPO ambitions | Largest-ever IPO could value a frontier lab near $2T | Public markets are being asked to price frontier AI risk at unprecedented scale |
| Streamers sue Twitch/Amazon over AI training | Creator-content training faces its biggest legal test | How platforms handle consent and compensation for training data is being defined in court |
| MAHA vs coal-powered data centers | Health advocates collide with AI energy demand | The environmental politics of the AI buildout are moving to the center of policy |
| AI agents vs Resy | Agents hit platform anti-bot defenses head-on | The agentic web is creating new rules of engagement for automated actors |
| UK sovereign AI bets on chip newcomers | Governments seek compute independence | National AI strategies are becoming a competitive force in the chip market |
| AI trained on living skin | Training data expands into live biological systems | Next-generation AI training is moving beyond internet text and images |
| OpenAI builds in-house economics research | Labs staff up to analyze their own impact | Economic and policy research is becoming core strategy for frontier labs |
What to Watch
- Anthropic’s S-1: With the NYT reporting a possible $100B raise and a $2T target, the actual filing will be the single most scrutinized document in AI this year — watch for the risk-factor section.
- Twitch/Amazon suit: The streamer lawsuit could force platforms to formalize consent and licensing for creator content used in AI training; watch for discovery requests and a possible class certification.
- US data center energy policy: MAHA’s pushback on coal adds a health angle to the power fight; watch for executive actions or EPA signals on data center power sources.
- Agent-platform relations: Resy-style bans will multiply as agents scale; watch for platform API policies and anti-agent terms updates.
- UK chip strategy: Watch which domestic chip startups receive sovereign funding and whether Callosum-style workload-routing software gains enterprise traction.