Politico Tests AI Access Through Claude ChatGPT and Copilot

The United States and nine partners—Australia, India, Israel, Japan, South Korea, Qatar, Singapore, the United Arab Emirates and the United Kingdom—signed the San Francisco Declaration on November 15 under the Pax Silica initiative. It aims to protect the technologies and infrastructure needed for advanced AI, including semiconductors, critical minerals, energy, communications, transport, logistics and digital platforms. The declaration calls for trusted, diverse supply chains, fewer excessive dependencies and cooperation against technology theft, economic pressure and disruptions. Five working groups will address innovation policy, AI adoption, supply-chain cooperation, security hubs and resilience. Germany, Sweden, the Netherlands and the European Commission are expected to sign next week, according to reports.

AI oversight is also emerging as a priority in Congress. Congressional Democrats are expected to push for stronger scrutiny of consumer risks, fairness, safety and transparency, as well as data-center power costs and cybersecurity, in contrast with the Trump administration’s focus on reducing regulation. Even without new laws, companies may face more investigations and pressure to demonstrate safeguards.

Luxor Technology says hackers attacked its Commander platform on November 14, changing settings to redirect customers’ Bitcoin mining power. Just over a dozen accounts were affected, and Luxor patched the flaw within two hours. Reports of roughly 64,000 redirected machines remain unconfirmed. Two payouts totaling about 0.162 BTC, worth approximately $10,500, went to an address researchers linked to the attack. Luxor says it will compensate affected customers and publish a technical report. CEO Nick Hansen said attackers used an AI-enabled method but did not explain how.

Politico is testing a service that lets a small group of Politico Pro Energy subscribers use its reporting through Claude, ChatGPT and Microsoft Copilot. The beta, running through early next year in the United States and Europe, supports questions such as requests for a briefing on data-center electricity demand, with answers linking to Politico stories. Subscribers can also set up automated agents; access follows each organization’s subscription, and Politico says its content will not train AI models.

Several other developments focus on putting AI to work. Amazon Bedrock’s prompt caching reuses eligible context—such as system prompts, documents, tool definitions and conversation history—to cut input costs and response times. Users can rely on automatic implicit caching or set explicit breakpoints, and usage data shows how much content came from cache. Cloudflare launched Kitesurf, a lightweight browser for AI agents. Its benchmarks show 3.1 to 3.8 times less processing power and up to seven times less memory use than Chromium, though some tasks are slower; it cannot play video, maintain long-lived logins or pass some bot checks. Cloudflare says automated systems account for 57.4% of web requests worldwide and plans to open-source Kitesurf.

Arist CEO Michael Ioffe says AI agents could recommend major company decisions, including layoffs, within the next few years, while executives retain final authority. Arist’s agents analyze employee interviews and company data, and the company works with more than 15% of the Fortune 500. Ioffe says the tools could surface answers in days rather than months of consulting, but acknowledges employees may hesitate to speak openly to an employer’s AI. Separately, researchers testing whether LLMs can manage behavioral clusters found their selected approach achieved the highest F1 score and required fewer interventions across YooChoose, MOOC and News-Category datasets. They caution that this is an observation from three datasets, not a general rule.

Other articles point to the breadth of AI’s financial and public impact. One describes Nvidia’s AI boom as tying Americans to the sector and cites a recent ousting of CEO Jensen Huang, alongside a data center under construction in Vernon, California; it provides no further detail on the reported leadership change. Mubadala consumer head Andrew Wang called AI an “incredible opportunity” for consumer private equity, pointing to the fund’s purchase of Chinese coffee company Luckin.

Key Takeaways

  • The United States and nine partners signed the November 15 San Francisco Declaration to secure AI-related supply chains and infrastructure through five planned working groups.
  • Germany, Sweden, the Netherlands and the European Commission are expected to sign the declaration next week, according to reports.
  • Congressional Democrats are expected to prioritize stronger AI oversight, while the Trump administration favors reducing regulation.
  • Luxor patched a Commander platform flaw within two hours; just over a dozen customer accounts were affected, while reports of about 64,000 redirected machines remain unconfirmed.
  • Politico is testing access to its energy reporting through Claude, ChatGPT and Microsoft Copilot, and says its content will not be used to train AI models.
  • Amazon Bedrock prompt caching can reuse context to lower input costs and response times, with implicit and user-configured explicit caching options.
  • Cloudflare’s Kitesurf browser uses less processing power and memory than Chromium in company benchmarks, but lacks some browser capabilities; the company plans to open-source it.
  • Arist CEO Michael Ioffe says AI agents could recommend decisions such as layoffs in coming years, with executives retaining final authority.
  • A study found an LLM approach performed best on three behavioral-clustering datasets, while researchers cautioned against treating the result as a general rule.
  • Mubadala’s Andrew Wang described AI as a major consumer private-equity opportunity; a separate article links Nvidia’s AI boom to Americans’ financial exposure to the sector.

India joins US led pact to protect AI supply chains

India joined the United States and several other countries in signing the San Francisco Declaration on the Secure Foundations of Enduring Prosperity, according to the Financial Times. The declaration aims to protect critical technologies and artificial intelligence supply chains as competition with China grows. It is part of Pax Silica, a US led effort focused on supply chains for semiconductors, critical minerals, energy and advanced manufacturing.

India and nine other nations pledge to secure AI supply chains

India, the United States and eight other countries signed the San Francisco Declaration, calling the protection of artificial intelligence supply chains a shared effort amid competition with China. The agreement was signed at a meeting of the US led Pax Silica initiative, convened by US Under Secretary of State for Economic Affairs Jacob Helberg. It says economic security is part of national security and proposes five working groups on innovation policies, technology adoption, supply chain cooperation, security hubs and resilience. The declaration also says that countries must work together to protect the supply chains and infrastructure needed for what the US government calls Super Intelligence.

US and nine partners sign pact to protect AI supply chains

The United States and nine partner countries signed the San Francisco Declaration on the Secure Foundations of Enduring Prosperity to protect technologies and supply chains needed for advanced artificial intelligence. The signatories are Australia, India, Israel, Japan, South Korea, Qatar, Singapore, the United Arab Emirates, the United Kingdom and the United States. The agreement calls for trusted and varied supply chains and working groups on innovation, AI adoption, supply chain cooperation and security. Germany, Sweden, the Netherlands and the European Commission are expected to sign next week, according to the Financial Times. The agreement comes before a planned meeting involving US President Donald Trump, Chinese President Xi Jinping and Russian President Vladimir Putin at the APEC Economic Leaders’ Meeting in Shenzhen.

US and nine partners sign declaration on AI security

The United States and nine partner countries signed the San Francisco Declaration on November 15 at the Pax Silica Founders Retreat, convened by US Under Secretary of State for Economic Affairs Jacob Helberg. The agreement aims to protect technologies and infrastructure that support super intelligence supply chains. Its five planned working groups will focus on innovation policies, adoption in key sectors, supply chain coordination, security hubs and resilience. The declaration calls for more trusted and varied supply chains and cooperation against economic pressure, supply disruptions and policies that distort markets. It also identifies sensitive technologies and infrastructure such as ports, transport networks, communications systems, energy facilities and digital platforms.

South Korea joins US led pact to protect AI supply chains

South Korea joined the San Francisco Declaration, a US led pact to protect critical artificial intelligence technologies and supply chains from hostile actors, the government announced on Wednesday. Korea and nine other countries committed to preventing the theft of sensitive technology and intellectual property that could threaten national security or disrupt global trade. The pact is part of wider efforts by the United States and its allies to defend against cyberattacks and other malicious activity. The article describes the agreement as a step in the US effort to rally allies before talks between US President Donald Trump and Chinese President Xi Jinping.

Israel joins pact to protect critical AI technologies

Israel and other Pax Silica members signed the San Francisco Declaration on November 15. The agreement sets out ways to protect advanced technologies and critical infrastructure and improve the resilience of participating countries. It calls for cooperation across the AI supply chain, from hardware and software to energy, communications, logistics and data. Israel presented a model from the National Institute for Impact-Driven Artificial Intelligence, and National AI Directorate head Erez Askal said strong alliances help protect technological independence.

US and nine partners sign pact to secure AI supply chains

The United States and nine partners signed the San Francisco Declaration, reported by Reuters on November 16. The countries are Australia, India, Japan, South Korea, Qatar, Singapore, the United Arab Emirates, the United Kingdom and Israel. The agreement, part of the US-led Pax Silica initiative, sets up five working groups on innovation, AI use in key industries, supply chain coordination and security, and broader resilience issues. It aims to protect sensitive technology and infrastructure and secure supplies of critical minerals, semiconductors and other technologies needed for AI.

US and nine partners agree to secure AI supply chains

The United States and nine partner countries signed the San Francisco Declaration on Sunday under the US-led Pax Silica initiative. The agreement aims to protect sensitive technologies and infrastructure needed for advanced AI, including transport networks, digital platforms, energy systems and logistics. It calls for trusted, diverse supply chains and fewer excessive dependencies. Germany, Sweden, the Netherlands and the European Commission are expected to sign next week, according to the Financial Times, citing US Under-Secretary of State Jacob Helberg.

US and nine allies sign declaration to protect AI technology

The United States and nine allies signed the San Francisco Declaration, according to a Reuters report. The agreement aims to coordinate efforts to protect AI technologies and critical infrastructure.

AI regulation set to top congressional Democrats’ priorities

Regulating artificial intelligence will be a major priority for congressional Democrats in the new year. The incoming Congress will face strong opposition from the Trump administration, which takes a hands-off approach to AI.

Election could bring tougher AI oversight and technology scrutiny

Democratic control of Congress is likely to shift AI policy toward stronger oversight of risks to consumers, fairness, safety and transparency. Lawmakers may also examine data center power costs and cybersecurity. Even if Congress does not pass new laws, businesses should prepare for more investigations and show that they have strong safeguards. This approach differs from the Trump administration’s focus on reducing regulation.

AI enabled attack redirects Bitcoin mining power from Luxor customers

Luxor Technology said hackers attacked its Commander platform on November 14, changing settings to redirect customers’ Bitcoin mining power to another pool. The company said just over a dozen customer accounts were affected, and engineers patched the flaw within two hours. Preliminary reports suggested about 64,000 machines may have been redirected, but Luxor has not confirmed that figure. Two payouts totaling about 0.162 BTC, worth roughly $10,500, went to an address linked by CoreX researchers to the attack. Luxor promised to compensate affected customers and publish a technical report, while CEO Nick Hansen said attackers used an AI enabled method but did not explain how AI was involved.

Politico tests access to energy reporting through AI tools

Politico is testing a service that lets some Politico Pro Energy subscribers use its reporting inside Claude, ChatGPT and Microsoft Copilot. A small group in the United States and Europe is taking part in the beta, which runs through early next year. Subscribers can ask questions, such as requests for a briefing on data center electricity demand, and answers link to specific Politico stories. They can also set up automated agents, while access follows each organization’s subscription. Politico says its content will not be used to train AI models.

Study tests LLMs for managing changes in behavioral clusters

Researchers tested whether a large language model can decide when to split, merge, retire or create behavioral clusters without re-clustering the data. Their approach uses statistics when cluster labels offer little detail and combines statistics with meaning when labels are more informative. Across three datasets, YooChoose, MOOC and News-Category, the selected LLM approach had the highest F1 score and needed fewer interventions than the comparison methods. Its advantage came more from precision than recall, and the useful information source changed across datasets. The researchers describe this finding as a practical observation from three datasets, not a proven general rule.

Nvidia’s AI boom shows how deeply Americans are invested

The article says Nvidia’s upheaval highlights how much Americans are tied to the AI boom, whether they want to be or not. It points to the recent ousting of Nvidia CEO Jensen Huang as an example. The article also describes a data center under construction in Vernon, California, photographed on October 29, 2026.

Mubadala sees AI as a major opportunity in consumer investing

Mubadala’s consumer head, Andrew Wang, called AI an “incredible opportunity” for consumer private equity. He pointed to the fund’s recent purchase of Chinese coffee company Luckin as a moment when the opportunity became clear. Wang shared his views with PEI Private Equity.

Amazon Bedrock prompt caching cuts costs and speeds up AI responses

Amazon Bedrock lets open-weight models reuse repeated context to lower input costs and response times. The article explains implicit caching, which reuses eligible context automatically, and explicit caching, which lets users set breakpoints for reusable parts. The methods can store system prompts, reference documents, tool definitions, and conversation history. Response usage data shows how much content came from cache, and examples include a support assistant and an OpenCode coding agent session.

Cloudflare builds a lightweight browser for AI agents

Cloudflare launched Kitesurf, a browser designed for AI agents, as automated systems make 57.4% of web requests worldwide and 68.6% in North America. Company benchmarks show Kitesurf uses 3.1 to 3.8 times less processing power and up to seven times less memory than Chromium, but some tasks run more slowly. It cannot play video, keep accounts logged in for long, or pass some bot checks, and Cloudflare says Chromium still handles those tasks. Nearly 132 million U.S. adults have shopped with AI assistance, but only 23% of merchants can clearly identify both AI traffic and the purchases it leads to. Cloudflare plans to open-source Kitesurf so customers can run their own copies.

Arist CEO says AI agents may recommend layoffs and other major decisions

Arist cofounder and CEO Michael Ioffe predicts that AI agents could recommend major company decisions, including layoffs, within the next few years, while executives keep final authority. Arist uses agents to interview employees and examine company data to identify business problems, and the company works with more than 15% of the Fortune 500. Ioffe says AI could help companies find answers in days instead of spending months and millions of dollars on consultants. Arist’s agents consider employee tenure, performance, and past engagement scores, and Ioffe says they can challenge executives who frame questions to get a preferred answer. He also acknowledged concerns about whether employees will speak openly to AI working for their employer.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI supply chains Pax Silica AI infrastructure semiconductors critical minerals AI regulation AI cybersecurity Bitcoin mining AI-powered cyberattacks AI journalism Claude ChatGPT Microsoft Copilot Amazon Bedrock prompt caching Cloudflare Kitesurf AI agents large language models AI in private equity Nvidia

Comments

Loading...