Anthropic is privately circulating a draft federal bill that would narrow when government can block a frontier model and preempt some state safety laws, while more than 20 lawmakers ask GAO to audit FAA's AI air-traffic tools a month after pilots began. Corporate AI adoption shows a widening gap between productivity claims and revenue, with UK research finding 77% of AI-using firms report no revenue change and Schwab shutting out unauthorized AI agents. Startup capital is clustering around defence and deep tech, from a $15M Ukrainian dual-use fund to NetApp and Sopra Steria corporate programmes.
- Sources read
- 15
- Claims that stand
- 8
- Quotes found in the source
- 16 of 16
- Claims struck
- 0
- Corrected or withdrawn
- 0
We quote it or we drop it. Every claim below carries the source’s own words, checked against the text we read.
Since 16 Oct
What we told you last time, ruled on from today’s evidence.
- Gone quietNothing new today on these. A limit of what we read, not a verdict.
AI data-centre financing stalls as CoreWeave debt nears distressed levels · ASML raises guidance a third time as customers eye 2029-2030 orders · Mount Sinai: brief safety prompt cuts harmful AI clinical choices to 10.1% · Eight agentic martech tools ship in 14 days from four vendors · Cyfendo pitches AI defence to small businesses after six-hour credential theft · Snabbit posts Rs 168 crore loss on Rs 12 crore revenue · AI drives new wave of Baltic unicorns, Estonia counts more than 10 · Chinese tech startups push into Japan at CEATEC despite tensions
Each mark on a claim’s line is an outlet, placed by when it published: 07:15 to 07:02 UTC.
AI
- Anthropic quietly pitches Congress a frontier AI safety bill with preemption
POLITICO obtained a confidential Anthropic draft bill that would let government block a model only on foreseeable, imminent risk of $1bn-plus damages or 50-plus deaths, with Commerce-registered evaluators and preemption of some state laws. The leading safety-branded lab is now writing the federal rules for its own products, and preemption would undercut the state laws it previously championed.
Representatives for the artificial intelligence giant Anthropic have quietly pitched Capitol Hill lawmakers on a draft bill to regulate frontier AI safety and security
2 more receipts
The bill would let the government block a model’s release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages
The draft proposal would limit the circumstances in which the government could block the release of an advanced model and preempt some state AI safety laws.
- Over 20 lawmakers ask GAO to audit FAA's AI air-traffic software
House Democrats requested a GAO review of FAA's FMDS and SMART platforms a month after pilots began over the National Capital Region, following an earlier letter asking for a pause. It is the first congressional oversight push on AI inside live safety-critical infrastructure, and the controllers' union is already drawing a line on replacement.
Congress is calling on the Government Accountability Office to audit the FAA’s AI-enabled air traffic software, only a month after the agency piloted the platforms.
1 more receipt
Over 20 members of Congress signed a letter to GAO asking for a review the FAA’s Flow Management Data and Services (FMDS)
- Microsoft's bias-correction framework wins sub-seasonal forecasting competition
Microsoft Research's Probabilistic Bias Correction, which learns from existing models' errors rather than replacing them, took top honours in a global two-to-six-week forecasting contest and is deployed on Kenyan monsoon forecasts. The two-to-six-week window is the gap that matters for agriculture, water and disaster planning, and the FT flags growing unease about leaning on AI for climate bets.
Probabilistic Bias Correction (PBC) is a new machine-learning framework developed by Microsoft Research that learns from a weather model’s past mistakes and helps correct them
2 more receipts
In early testing, PBC took top honors in a global competition aimed at improving extended-range, or sub-seasonal forecasting.
The technology is being used to anticipate everything from forest fires to futures contracts. Some fear that's a mistake.
AI agents that find customers for small businesses and founders
- Schwab blocks unauthorized AI agents, steers clients to in-house assistant
Charles Schwab is blocking third-party AI agents from acting on client accounts and directing users to its own 'Charley' assistant. Agent builders selling into financial customers face platform lockouts; incumbents will own the agent layer wherever they control the account.
Schwab blocks unauthorized AI agents, points clients to 'Charley'
- UK data: 77% of AI-using firms see no revenue change
CNBC reports Paysend's CEO rations AI tools to proven performers and claims 25% more engineering output, while UK government research finds productivity gains rarely reach the top line. Founders buying AI for growth should note the gap between hours saved and revenue earned before paying for seat-wide rollouts.
U.K. government research found 75% of AI-using businesses reported improved workforce productivity, but 77% reported no revenue change.
2 more receipts
Paysend CEO Ben Chisell limits access to general-purpose AI tools to high-performing employees who can show what the tools add.
estimates that Paysend can now deliver around 25% more engineering work with the same-sized team
tech startups
- WildField Ventures launches $15M Ukrainian fund for dual-use startups
A new Ukrainian VC fund targets up to $15 million for early-stage dual-use technology companies, as Pittsburgh startups court national-security investors at the AUSA meeting. Defence-adjacent capital is forming at seed stage on both sides of the Atlantic, giving founders a non-consumer exit path.
WildField Ventures, a new VC fund, has launched in Ukraine with a target of up to $15 million to back early-stage dual-use technology startups.
1 more receipt
Three Pittsburgh startups connected with national security investors at the annual Association of the United States Army meeting in Washington, D.C.
- Sopra Steria Ventures to link 80 French Tech 2030 startups with group
Sopra Steria's venture arm will connect the 80 French Tech 2030 companies with its business units and customers. Corporate distribution for a state-picked cohort is the kind of channel European startups usually lack.
The initiative will aim to create practical collaborations between the 80 startups and Sopra Steria's business units and customers.
- NetApp picks five deep-tech startups for 15th Excellerator cohort
NetApp's India accelerator selected five startups in AI software engineering, cybersecurity and workflow automation. Infrastructure incumbents keep sourcing agentic and security startups through accelerators, signalling where enterprise buyers expect to spend.
NetApp has selected five deep-tech startups for its 15th Excellerator cohort, focusing on AI software engineering, cybersecurity, workflow automation
Why it matters
The regulatory centre of gravity is shifting from states to Washington, and the labs are moving first: Anthropic's draft sets a $1bn/50-death threshold for intervention and would preempt state laws, which means the standard that California set could become the ceiling rather than the floor. At the same time Congress is discovering AI already running in air-traffic flow management, and the oversight request lands after deployment, not before.
For founders, the commercial read is sober. UK data showing productivity gains without revenue gains, a CEO gating tools to top performers, and Schwab walling off third-party agents all point the same way: AI agents that find or serve customers will have to prove revenue lift and win platform access, not just ship. Capital is still flowing, but toward defence, deep tech and corporate-sponsored cohorts rather than generic AI tooling.
What to watch
- Whether any House or Senate sponsor formally introduces Anthropic's draft and how state attorneys general respond to the preemption clause.
- GAO's response to the FAA audit request and whether DOT pauses FMDS/SMART over the National Capital Region.
- Whether other brokerages and banks follow Schwab in blocking unauthorized AI agents, closing the account layer to third-party agent startups.
