Anthropic designated supply chain risk as OpenAI follows suit

Key Takeaways

• The Pentagon officially designated U.S. AI firm Anthropic as a "supply chain risk," a first for an American company, due to disagreements over AI usage safeguards.
• Anthropic sought restrictions to prevent its Claude AI from being used for mass surveillance or autonomous weapons, while the Pentagon insisted on using AI for "all lawful purposes."
• Despite the "supply chain risk" label, the U.S. military continues to use Anthropic's Claude AI in operations, including against Iran.
• Talks between the Pentagon and Anthropic have reportedly resumed after President Trump's public criticism and the initial breakdown in negotiations.
• Businesses are adopting five AI value models, including workforce empowerment and AI-native distribution, to drive reinvention and enhance customer engagement.
• Banks face a critical need to scale generative AI responsibly, requiring significant investment in data, operating models, and workforce transformation.
• New smart glasses, like Halo, are moving AI processing off the cloud to the device itself, enhancing user privacy and reducing latency, challenging models from companies like Meta and Snap.
• A proposed New York bill seeks to hold AI platforms liable if chatbots impersonate professionals and provide incorrect advice, allowing users to sue.
• IRONSCALES has launched AI agents to proactively fight phishing attacks by generating realistic simulations and providing rapid forensic investigations.
• OpenAI is providing resources to help students and educational institutions overcome the "capability overhang" with tools like ChatGPT, promoting deeper and more effective AI applications.

Pentagon feuds with AI firm Anthropic over security concerns

The Pentagon has labeled AI company Anthropic a supply chain risk, creating confusion and unresolved questions. Talks between the two sides failed over how Anthropic's AI models could be used by the Department of Defense. Anthropic was previously the only AI company allowed on the Pentagon's classified networks. Experts find the situation puzzling, especially since the military continues to use Anthropic's AI for operations, even in sensitive areas like Iran. The Pentagon has not clearly explained the specific threat Anthropic poses.

Trump's AI agenda at risk amid Anthropic dispute

Critics argue that President Trump's administration is harming its own AI agenda by clashing with the American AI company Anthropic. Lobbyists and former officials warn that labeling Anthropic a supply-chain risk, a term usually for foreign adversaries, creates uncertainty and could make U.S. AI firms less competitive globally. This dispute may slow innovation and cede ground to China. The administration's actions are seen as a heavy-handed attack on the private sector, contradicting its pro-innovation goals.

Pentagon officially labels Anthropic a security risk

The U.S. military has officially designated the AI firm Anthropic as a supply chain risk, potentially cutting it off from contracts. This follows a disagreement over Anthropic's demand for safeguards preventing the military from using its Claude AI for mass surveillance or autonomous weapons. The Pentagon insists it needs AI for 'all lawful purposes' and that existing policies already cover these concerns. Despite the designation, the military is still using Claude for operations in Iran.

Pentagon confirms Anthropic is a supply chain risk

The Pentagon has officially informed Anthropic that its AI products are considered a risk to the U.S. supply chain. This designation, effective immediately, comes after talks broke down over Anthropic's demand for restrictions on AI use for mass surveillance or autonomous weapons. Despite the declaration, Anthropic's Claude AI tools are still being used by the U.S. military in operations against Iran. The Pentagon stated it needs to use technology for 'all lawful purposes' without vendor-imposed restrictions.

Trump fires Anthropic as Pentagon blacklists AI firm

President Trump stated he fired Anthropic 'like dogs' amid a dispute over AI usage. The Pentagon has now formally designated Anthropic a 'supply chain risk,' preventing government contractors from using its technology. This designation has never been applied to a U.S. company before. Reports suggest talks may have restarted between the Pentagon and Anthropic regarding the military's use of the company's AI. The conflict began when Anthropic refused a deal over concerns about its AI being used for surveillance or autonomous weapons.

Pentagon and Anthropic resume AI talks amid dispute

The Pentagon has reopened negotiations with AI company Anthropic less than a week after threatening to blacklist it. This comes as Anthropic's CEO, Dario Amodei, suggested the company was targeted for not donating to Trump or praising him. The Pentagon previously designated Anthropic a supply chain risk due to disagreements over safeguards for its AI models, which Anthropic wants to prevent from being used for mass surveillance or autonomous weapons. The Pentagon argues it needs AI for 'any lawful use' and points to a similar deal with OpenAI as a compromise.

Five AI value models for business reinvention

Businesses can reinvent themselves using five key AI value models, moving beyond simple use cases. These models, including workforce empowerment and AI-native distribution, create value differently and build upon each other. Workforce empowerment quickly spreads AI skills, building fluency for deeper transformation. AI-native distribution changes how customers discover and choose products through conversational engagement. Other models focus on expert capabilities, system integration, and agent-led operations, helping companies scale AI effectively.

EY: AI is part of a larger banking transformation

Banks are at a critical point with artificial intelligence, needing to scale generative AI (GenAI) responsibly. The main challenge involves significant change and investment in data, operating models, controls, and workforce. EY emphasizes that AI is just one piece of a broader transformation agenda. Winning banks will integrate AI with modernized technology, redesigned workflows, and better-equipped employees. Moving beyond experimentation requires discipline and integration, focusing on foundational modernization, workforce skills, culture, and process redesign.

Reform deputy defends AI photo as political art

Reform UK deputy leader Darren Grimes is defending his use of an AI-generated image depicting him in front of a burning County Durham. Critics have called the image 'fake news,' but Grimes stated it was clearly labeled as AI-generated and intended as political commentary on regional challenges. He urged critics to focus on real issues rather than the AI artwork. The image has sparked debate about the use of AI in political messaging and the spread of misinformation.

AI projects makeup onto faces based on descriptions

Researchers have developed an AI system that projects makeup colors onto a user's face based on spoken descriptions of moods or styles. This technology learns user preferences in real time and displays results under realistic lighting, offering a more lifelike experience than traditional virtual makeup apps. Users can describe concepts like 'Sakura in spring,' and the AI generates personalized makeup color suggestions for cheeks, eyeshadow, and lips. The system uses a projector to apply colors directly to the face, adapting to skin tone and texture.

New smart glasses move AI processing off the cloud

Brilliant Labs, Neuphonic, and TheStage AI have partnered to create smart glasses, called Halo, that process AI directly on the device instead of relying on the cloud. This approach aims to reduce latency and enhance user privacy by keeping sensitive data local. The Halo glasses will feature on-device vision inference and Neuphonic's conversational AI models, optimized by TheStage AI's engine. This challenges cloud-dependent models from companies like Meta and Snap by prioritizing user privacy and faster response times.

New York bill targets AI impersonating professionals

A proposed law in New York aims to prevent AI chatbots from impersonating lawyers, doctors, and other licensed professionals. The bill would allow users who rely on incorrect advice from such AI to sue the platforms. Companies would not be able to avoid liability by simply notifying users they are interacting with a chatbot. This legislation is part of broader efforts in New York to regulate AI, including protections for minors and requirements for AI platforms.

Blueprint for securing autonomous AI agents

Securing autonomous AI agents requires a multi-layered approach, similar to defense-in-depth strategies. Developers should adopt an adversarial threat model early in the design process, using principles like sandboxing code execution and enforcing role separation in multi-agent systems. Limiting an agent's scope, granting minimum tool access, and implementing human-in-the-loop oversight for sensitive actions are crucial. A real-time defensive layer using AI to monitor agents and offensive testing through automated red teaming are also recommended to build trust and prevent manipulation.

Four reasons AI may not take your job

Despite rapid AI advancements, mass white-collar unemployment is not a certainty. One reason is that AI adoption and job loss haven't significantly correlated yet, with unemployment rates remaining stable. Companies are still hiring, even while investing in AI. Furthermore, the evidence for exponential AI progress has methodological flaws, and economic factors like interest rate hikes may better explain recent hiring slowdowns. The argument suggests that AI's impact on jobs may be more nuanced than predicted.

IRONSCALES uses AI agents to fight phishing attacks

IRONSCALES has launched AI agents to help security teams combat increasingly sophisticated AI-driven phishing attacks. These agents continuously research an organization's public information to generate realistic phishing campaigns, hardening detection models before real attacks occur. The system creates a closed-loop process where reconnaissance feeds detection, which then informs training, all without human intervention. A Phishing SOC Agent also provides rapid, detailed forensic investigations of suspicious emails, reducing response times from hours to minutes.

OpenAI helps students close AI capability gaps

OpenAI is providing tools and resources to help educational institutions address the growing gap between AI capabilities and how people use them. College students, who are major users of ChatGPT, often operate far below the potential of these tools. OpenAI's research shows a significant 'capability overhang,' where students need deeper applications beyond basic tasks. By embedding authentic AI use cases into coursework, educators can help students develop agency and harness AI's full potential for future opportunities.

Sources

• 5 unresolved questions hanging over the Anthropic–Pentagon fracas: 'It's all very puzzling'
• 'Bitterly ironic’: Trump is wrecking his AI agenda with Anthropic spat, lobbyists and ex-officials say
• Pentagon formally designates Anthropic a supply chain risk amid feud over AI guardrails
• Pentagon Says It’s Told Anthropic the Firm Is Supply-Chain Risk
• Trump says he fired Anthropic ‘like dogs’ as negotiations with Pentagon reportedly restart
• Can Anthropic Make Peace With the Pentagon?
• The five AI value models driving business reinvention
• EY: AI Is Just One Piece of the Banking Transformation Puzzle
• Reform deputy Darren Grimes AI photo condemned as 'fake news'
• Describe the vibe, see the look: An AI-based system projects makeup onto the user's face
• Smart Glasses, Rebuilt for Privacy: Brilliant Labs, Neuphonic & TheStage AI Move AI Off the Cloud
• Proposed New York law would bar AI chatbots from posing as lawyers, allow duped users to sue
• Engineering trust: A security blueprint for autonomous AI agents
• 4 reasons why AI (probably) won’t take your job
• IRONSCALES Introduces AI Agents to Help Security Teams Address AI-Driven Phishing Attacks
• Ensuring AI use in education leads to opportunity