Anthropic has introduced FlashProve, an open-source tool that lets users verify which AI model generated a specific response. The system uses cryptographic proofs, similar to email signatures, to confirm authenticity without exposing model weights. Anthropic released the code on GitHub as a work in progress, noting that current users have no way to check if a model was secretly swapped by a routing service or attacker.
In the AI policy arena, Leading the Future, a pro-AI super PAC backed by Silicon Valley billionaires, spent $93.6 million on midterm elections through ads and lobbying. Despite its roughly $100 million war chest, many candidates it supported lost their races. The group focused heavily on opposing AI regulation, which may have limited its broader appeal. Fellow AI PACs American Mission and Building the Future also spent $4.8 million backing South Carolina Senate candidate Darline Graham, who likewise lost.
Financial institutions are moving quickly on AI adoption but caution about responsibility. State Street executive Manoj Bohra said the bank has over 350 AI use cases, up from about 150 last year, and stressed that speed cannot come at the cost of safety. Executives from Goldman Sachs and Morgan Stanley discussed the emerging era of agent-to-agent banking at the Global Financial Leaders Investment Summit in Hong Kong.
On the education and enterprise front, Great Lakes AI Week runs November 10 to 12 at Bowling Green State University and The University of Toledo, featuring speakers and employers like Capital One, Salesforce, and Microsoft Copilot. BRI Wealth Management adopted Aveni Assist and cut report writing time by 75 percent, letting advisers focus more on portfolio management and client service. Andavi Solutions acquired Infilect to add visual AI tools that analyze store photos for shelf checks, pricing, and competitor activity.
Microsoft faces headwinds with its AI PC push, as high memory costs from data centers have raised PC prices and demand remains slow. Noah Shinn, 24, founded Instinct to build AI shopping agents, speaking about the project at Fireworks AI's Forge conference in San Francisco. Supermicro reported record AI server sales while expanding into full data center systems, though questions remain about lasting profitability. Hugging Face CEO Clément Delangue said open source models will build AI trust, arguing that closed source models make it hard for users to verify results or understand decisions.
Key Takeaways
- Anthropic launched FlashProve, an open-source tool using cryptographic proofs to verify which AI model generated a response.
- Pro-AI PAC Leading the Future spent $93.6 million on midterms but many supported candidates lost.
- AI PACs American Mission and Building the Future spent $4.8 million backing Darline Graham, who also lost.
- State Street has over 350 AI use cases, up from about 150 last year, and emphasizes responsible deployment.
- Great Lakes AI Week runs November 10-12 at Bowling Green State University and The University of Toledo, featuring Salesforce, Capital One, and Microsoft Copilot.
- BRI Wealth Management cut report times by 75% using Aveni Assist while keeping human advisers in charge of final recommendations.
- Microsoft struggles with slow demand and rising costs for AI PCs despite pushing Copilot adoption.
- Andavi Solutions acquired Infilect to add visual AI tools for retail shelf and pricing analysis.
- Supermicro reported record AI server sales but profitability remains uncertain.
- Hugging Face CEO Clément Delangue says open source models will build public trust in AI by enabling transparency.
Anthropic launches FlashProve to verify AI model identity
Anthropic announced FlashProve, an open-source tool that lets users confirm which AI model generated a response. It uses cryptographic proofs similar to email signatures to verify authenticity. The system protects model weights while allowing independent verification. Anthropic released the code on GitHub as a work in progress.
Anthropic builds tools to confirm which AI model gave an answer
Anthropic developed FlashProve, a system designed to deliver provable inference so users can confirm responses came from a specific AI model. The company said current users cannot check if a model was secretly substituted by a routing service or attacker. Cryptographic proofs let a provider commit to a model and prove each response came from it. Anthropic said this has implications for security and customer trust.
Pro-AI super PAC Leading the Future failed to sway midterms
Leading the Future, backed by Silicon Valley billionaires, spent $93.6 million on midterm elections through ads and lobbying. Despite its large war chest of about $100 million, many candidates it supported lost their races. The group focused heavily on opposing AI regulation, which may have limited its appeal to broader voters. Leader Joe Lonsdale said the group is not giving up.
AI PACs spent millions on midterms with mixed results
AI-focused political action committees spent millions trying to influence midterm elections and shape AI regulation. American Mission and Building the Future spent $4.8 million backing South Carolina Senate candidate Darline Graham, who lost. Other candidates supported by these groups also lost races. Pro-regulation groups also saw mixed results, and both sides plan to keep spending in future elections.
Banks warn of AI safety risks as agentic finance era begins
State Street executive Manoj Bohra spoke at the Global Financial Leaders Investment Summit in Hong Kong about AI in banking. He said State Street has over 350 AI use cases, up from about 150 last year. He stressed that speed in deploying AI cannot come at the cost of responsibility. Executives from Goldman Sachs, Morgan Stanley and other firms discussed the dawning era of agent-to-agent banking.
Great Lakes AI Week Brings Leaders to UToledo in November
Great Lakes AI Week runs November 10 to 12 at Bowling Green State University and The University of Toledo. Speakers and employers include Capital One, Salesforce, Cintas, IMEG and Nexigen. The event offers free AI credentials through AI Ready Ohio. Attendees can join workshops on tools like Microsoft Copilot and Gemini Notebook.
BRI Wealth Cuts Report Times by 75% Using Aveni AI
BRI Wealth Management adopted Aveni Assist in 2025 and expanded it in 2026. The AI tool handles repetitive report writing, cutting report time by 75 percent. Staff now focus more on portfolio management and client service. BRI says advisers still make all final recommendations.
Windows AI PCs Face Rising RAM Prices and Slow Demand
Microsoft is urging companies to refresh PCs with AI features and Copilot. High memory costs from AI data centers have raised PC prices. Many customers are delaying purchases because of the cost and few must-have apps. Microsoft wants Windows to become a platform for running AI agents locally.
Andavi Solutions Acquires Infilect for AI Retail Tools
Andavi Solutions bought Infilect to add visual AI tools to its retail platform. Infilect uses image recognition and Agentic AI to analyze store photos and videos. The technology checks shelves, pricing, availability and competitor activity. Retail teams can get next-best actions during store visits.
Instinct Founder Noah Shinn Talks AI Agents for Shopping
Noah Shinn, 24, founded Instinct to build AI agents that help people shop and spend. He spoke about the project at Fireworks AI's Forge conference in San Francisco. Shinn says people quickly trust AI with their money. Instinct focuses on making AI shopping tools fast and easy to use.
Supermicro AI server sales grow but profits remain uncertain
Supermicro reported record sales for its AI server products as it expands into full AI data center systems. The company is now focusing on whether this growth will lead to lasting profits and better cash flow. Details about the financial results require a paid subscription to read fully.
Hugging Face CEO says open source models will build AI trust
Clément Delangue, the CEO of Hugging Face, believes that open source models will help people trust artificial intelligence. He noted that Hugging Face was recently attacked by rogue AI agents created by OpenAI. Delangue argues that closed source models make it hard for users to verify results or understand how decisions are made. By sharing the source code and training data, companies can show exactly how their AI works. He thinks this transparency will change public perception from seeing AI as a threat to viewing it as a useful tool.
Sources
- What is FlashProve? Anthropic’s open-source tool to prove which AI model generated a response
- Anthropic Develops Tools to Confirm AI Models’ Identities
- The Biggest Pro-AI Super Pac Network Didn't Really Impact the Midterms
- AI PACs spent millions trying to shape Congress. Here's how they fared.
- Banks confront daunting new AI safety risks as era of agentic finance dawns
- Great Lakes AI Week Brings Heavy Hitters to UToledo on Nov. 12
- BRI Wealth cuts report times by 75% with Aveni AI
- Windows pushed AI PCs, but now it's hit a wall of RAM prices
- Andavi Solutions Acquires Infilect to Expand AI-Powered Retail Execution and Agentic AI Capabilities
- Instinct founder Noah Shinn touts AI agents racing to shop
- Supermicro's record AI server sales put margins and cash conversion in focus
- Open Source Models Will Convince People AI Isn’t Bad: Hugging Face CEO
Comments
Please log in to post a comment.