Microsoft supports Anthropic against Pentagon $200 million deal

Key Takeaways

• Microsoft is supporting AI company Anthropic in its legal challenge against the Pentagon's

  Microsoft backs Anthropic in Pentagon AI dispute

  Microsoft is supporting AI company Anthropic in its legal fight against the US Pentagon. Anthropic is challenging a designation that labels its AI as a supply-chain risk, which could prevent it from government work. Microsoft filed a court brief arguing the government needs access to the best technology while ensuring AI is not used for mass surveillance or without human control. This dispute arose after contract talks for a $200 million deal failed when Anthropic insisted its AI not be used for mass surveillance or autonomous weapons. The Pentagon then labeled Anthropic a risk, leading to cancelled contracts.

  Microsoft supports Anthropic against Pentagon AI ban

  Microsoft has filed a legal brief asking a court to stop the Pentagon from labeling Anthropic's AI technology a national security risk. This action follows a breakdown in contract negotiations where Anthropic refused to allow mass surveillance or fully autonomous weapons. The Pentagon then designated Anthropic a risk, prompting Microsoft's support. Microsoft stated that AI should not be used for domestic surveillance or to start wars without human control. The company also warned that the Pentagon's quick designation could disrupt military operations and cause wide-ranging costs for contractors.

  Microsoft joins retired military leaders backing Anthropic

  Microsoft and a group of retired military leaders are supporting AI company Anthropic in its legal challenge against the Pentagon. They are asking a federal court to block the Trump administration's decision to label Anthropic's AI products as a supply chain risk. This designation by Defense Secretary Pete Hegseth followed a public disagreement over Anthropic's refusal to allow unrestricted military use of its AI model Claude. Microsoft argued that using a supply chain risk designation for a contract dispute could harm the public interest and force contractors into compliance with unclear directions.

  Anthropic sues Pentagon after AI products blacklisted

  AI company Anthropic has asked a court to halt the Pentagon's decision to blacklist its products, citing potential irreparable harm to its business. The company filed a lawsuit in federal court in California and requested a stay from the U.S. Court of Appeals for the District of Columbia Circuit. Anthropic stated the Pentagon's actions could lead to billions of dollars in lost revenue for 2026. The dispute began over safeguards for military use of Anthropic's AI tools, specifically regarding mass surveillance and autonomous weapons.

  Anthropic seeks emergency halt to Pentagon AI designation

  Anthropic is requesting an emergency stay from the Pentagon's designation of its products as a supply chain risk. The company argues that Defense Secretary Pete Hegseth's decision will cause escalating and irreparable harm. Anthropic claims the designation violates due process as no legal or factual explanation was provided. This action follows a clash over the use of AI for mass surveillance and autonomous weapons, leading President Trump to order federal agencies to stop using Anthropic's products. Anthropic also alleges retaliation for its stance on AI safety.

  Anthropic: The AI company focused on safety and disruption

  AI company Anthropic is at the forefront of developing powerful AI while prioritizing safety. Recently, Anthropic delayed the release of its Claude 3.7 Sonnet model for 10 days after a trial indicated it could help terrorists create biological weapons. This incident highlights the company's commitment to addressing worst-case scenarios, even as they push the boundaries of AI technology. CEO Dario Amodei aims to balance AI's benefits with its risks, acknowledging the immense responsibility involved in developing such powerful tools.

  Anthropic CEO Dario Amodei's vision for AI's future

  Anthropic CEO Dario Amodei envisions a future where advanced AI, like his company's Claude, could solve major global problems and extend human lifespans. However, he also deeply worries about AI's potential dangers, comparing its development to the Manhattan Project. Amodei believes AI must be developed responsibly to avoid overpowering humanity or disrupting global power balances. His recent conflict with the Pentagon over AI use in warfare and surveillance suggests that controlling AI's application may be a complex challenge.

  Guide to implementing agentic AI in businesses

  Implementing agentic AI requires a shift in how work is defined and decisions are made, moving beyond simple software features. Many companies struggle with AI pilots that fail due to issues with data, governance, and unclear success metrics. This guide emphasizes that successful agentic AI operates like a well-run team, with each agent having a clear role, supervision, and a process for improvement. It highlights the need for detailed work definitions, bounded autonomy, and a culture of continuous improvement to bridge the gap between AI investment and tangible value.

  Journalists can use AI as source colleague or assistant

  Journalists can effectively use AI by categorizing its role as a source, colleague, or assistant. While AI can synthesize vast information as a source, it requires verification like any human source. As a colleague, AI can help refine ideas and provide different perspectives. As an assistant, AI can aid with tasks like outlining and fact-checking, but human oversight is crucial for nuanced writing and complex narratives. Newsrooms need clear guidelines for AI use to manage risks and leverage opportunities, despite initial resistance from some reporters.

  Pope Leo XIV's balanced view on AI development

  Pope Leo XIV's message for the World Day of Social Communications emphasizes a balanced perspective on Artificial Intelligence. He cautions against apocalyptic forecasts, drawing parallels to historical reactions to writing and the printing press. The Pope believes AI, like previous technologies, can be used for good or evil, and human responsibility is key. He warns against naive reliance on AI as an all-knowing source, stressing the need for discernment and control over its development and application.

  AI facial recognition error wrongly jailed grandmother

  Angela Lipps, a grandmother from Tennessee, spent nearly six months in jail due to an AI facial recognition error. Fargo police used the software to identify her as a suspect in a North Dakota bank fraud case. Lipps maintained her innocence, stating she had never been to North Dakota. Her attorney eventually proved her innocence by presenting bank records showing she was over 1,200 miles away when the fraud occurred. This case highlights the potential dangers of relying solely on facial recognition technology, as it has led to wrongful arrests in other instances.

  AI fuels record UK fraud cases

  A record 444,000 fraud cases were reported in the UK last year, driven by criminals using advanced AI tools. The UK's leading anti-fraud body, Cifas, warns that AI enables large-scale deception and account takeovers for mobile, banking, and online shopping. Criminals are using AI for sophisticated impersonation and creating synthetic identities. This trend is expected to continue, making identity fraud and account takeover significant threats. Fraud accounts for over 40% of all crime in the UK, with many consumers lacking confidence in spotting AI-enabled scams.

  Hacker Matt Suiche discusses AI's impact on cyberwar

  Legendary hacker Matt Suiche discusses the evolving landscape of cyberwarfare in the age of AI. He notes the blurring lines between physical and digital warfare, citing recent drone strikes on data centers and cyberattacks on infrastructure. Suiche, founder of data infrastructure startup OnDB, explores Iran's cyber capabilities and how AI is transforming coding and hacking. The conversation touches upon the increasing sophistication of digital warfare and the potential threats posed by AI-powered cyberattacks.

  Waton Financial and Tsinghua University launch AI Fintech Lab

  Waton Financial Limited, Tsinghua University's X-Tech, and PandaAI have partnered to establish the 'AI and Fintech Joint Lab'. This collaboration aims to explore the application of AI Agents in real-world trading scenarios, moving beyond AI as a simple tool to a decision-making partner. The lab will focus on equipping AI Agents with capabilities like understanding market sentiment, geopolitical factors, and executing trades autonomously. This initiative seeks to democratize quantitative trading, making advanced investment tools accessible to individual investors.

  New tool detects AI security risks in RAG systems

  A new open-source tool called Adversarial Hubness Detector has been developed to address security vulnerabilities in Retrieval-Augmented Generation (RAG) systems. These systems, commonly used in enterprise AI, can be susceptible to 'zero-click' poisoning where adversaries manipulate data embeddings. This can force malicious content into search results for thousands of unrelated queries. The tool uses multiple detectors to identify statistically anomalous data points, helping to flag these adversarial attractors before they impact users and ensuring the security of AI models.

  Sources

  • Microsoft backs AI firm Anthropic in legal battle against Pentagon
  • Microsoft sides with AI startup Anthropic in legal fight with Pentagon
  • Microsoft backs Anthropic, urging a judge to halt Pentagon’s actions against AI company
  • Anthropic Seeks Court Stay After Pentagon Blacklists AI Products
  • Anthropic requests emergency stay of supply chain risk designation in DC appeals case
  • The Most Disruptive Company in the World
  • The Dissonance of Anthropic's CEO, Dario Amodei
  • Operationalizing Agentic AI Part 1: A Stakeholder’s Guide
  • How journalists can make AI work for them.
  • Leo XIV’s balanced view of Artificial Intelligence
  • Tennessee grandmother jailed after AI facial recognition error links her to fraud
  • ‘Convincing’ AI scams drove UK fraud cases to record 444,000 last year
  • Legendary Hacker Matt Suiche on Cyberwar in the Age of AI
  • Waton Financial Limited Partners with Tsinghua University to Establish AI and Fintech Joint Lab
  • Your Model's Memory Has Been Compromised: Adversarial Hubness in RAG Systems