Anthropic launches Claude Tag, an always-on AI teammate for Slack

Title: AI Agents Learn Enterprise; Government Tightens Security

01Anthropic Launches Claude Tag, an Always-On AI Teammate for Slack

Anthropic has released Claude Tag, a new enterprise feature that embeds an AI agent directly into Slack to observe and learn organizational workflows in real-time. Unlike standalone AI assistants, Claude Tag is designed to accumulate institutional knowledge—understanding team dynamics, project histories, and communication patterns over time. The system marks a strategic shift from AI as a tool to AI as an ongoing participant in organizational processes.

Beyond productivity gains, the move represents Anthropic's bid to capture enterprise workflows and organizational context as a competitive moat. By embedding deeply into daily communications, Claude Tag can offer increasingly personalized assistance while giving Anthropic valuable data on how organizations actually operate. The feature enters a crowded market for workplace AI but differentiates through its persistent memory of company-specific information.

02GPT-5 Pro Helps Immunologist Solve Three-Year-Old T Cell Mystery

OpenAI's GPT-5 Pro has demonstrated practical scientific utility by helping immunologist Derya Unutmaz decode a three-year-old mystery about T cell behavior. The model analyzed complex immunological data and identified patterns that had eluded traditional research methods, offering new insights into how T cells function and respond to threats. The breakthrough carries implications for both cancer immunotherapy and autoimmune disease treatment.

The case adds to growing evidence that frontier AI models can accelerate scientific discovery in specialized domains. Unutmaz reported that GPT-5 Pro's ability to synthesize information across immunology, genetics, and cellular biology helped him form hypotheses that would have taken months to develop through conventional analysis. OpenAI highlighted the collaboration as a proof-of-concept for AI-assisted scientific research, though experts note that independent validation of findings remains essential.

03US Government Accelerates Deadline for Post-Quantum Cryptography Adoption

A new executive order has moved up the federal deadline for transitioning away from quantum-vulnerable encryption, citing heightened national security risks. The order warns that adversaries could be harvesting encrypted data now to decrypt later once sufficiently powerful quantum computers exist—a threat known as "harvest now, decrypt later" attacks. Federal agencies now face an accelerated timeline to adopt post-quantum cryptography standards.

The urgency reflects growing concern that current encryption methods will become obsolete as quantum computing advances. The order pushes agencies to prioritize systems handling sensitive data and critical infrastructure, with stricter requirements for agencies managing classified information. Security experts have long warned that the transition to quantum-resistant encryption requires years of planning, making early action essential to avoid a rushed, error-prone migration.

Also today

04Mistral releases OCR 4 with improved document understanding — The French AI lab's latest model offers significantly improved parsing for structured and unstructured documents.
05ASML's $400 million machine powers next-generation chipmaking — The Dutch company's extreme ultraviolet lithography machines remain critical for advanced AI chip production.
06Research reveals AI hiring tools can perpetuate racial bias — Stanford HAI study finds algorithmic hiring systems may systematically reject qualified candidates based on protected characteristics.
07AI's affordability crisis threatens sustainable development — Growing compute costs for training frontier models raise questions about long-term economic viability.
08Oracle cuts 21,000 jobs while investing billions in AI infrastructure — Workforce restructuring funds massive data center expansion for AI workloads.
09Omio builds AI-native travel platform with OpenAI — The European travel booking company leverages OpenAI models for conversational booking experiences.
10Fika Jobs raises $4M for AI agent interview platform — Stockholm startup combines AI interview agents with short-form video profiles for a LinkedIn-meets-TikTok hiring experience.