Regulation surges globally; Anthropic streamlines Claude

Anthropic / Claude ecosystem

Anthropic sets Claude Code to Auto Mode by default

Claude Code now defaults to Auto Mode, automatically approving certain actions to boost developer productivity while maintaining safety protections. The change aims to reduce developer friction from manual approvals during coding sessions.

Claude Code creator advises clearing configuration files every six months

The father of Claude Code recommends periodically resetting claude.md, skills, and hooks to let the model self-optimize. This practice aims to enhance Claude Code's performance over time.

Frontier model providers

No significant new developments.

AI developer tooling & infrastructure

Lybrary, a persistent AST-aware code memory for AI agents, is released

Lybrary enables AI agents to maintain an AST-aware memory of codebases across sessions, avoiding redundant re-reading. It functions as an MCP server for easy integration.

Cloud & platform providers

Google Cloud launches AI-powered security operations in Taiwan

Google Cloud's Security Operations platform is now available in its Taiwan region, offering AI-driven threat detection and response tailored to local data residency requirements.

Cloudflare shifts bot detection to continuous session analysis

Cloudflare is moving toward a self-adjusting bot-detection engine called Adaptive Intelligence for Bot Management, with broader rollout planned for enhanced security.

AI policy, regulation & governance

India demands Meta change algorithms to combat deepfakes and propaganda

In a regulatory escalation, India is pressuring Meta to modify its recommendation algorithms and enhance content moderation to curb deepfakes, propaganda, and harmful content.

Japan sets voice rights guidelines to target unauthorized AI use

Japan's Justice Ministry has released guidelines expanding publicity rights to protect the voices of celebrities and voice actors from unauthorized AI-generated deepfakes.

Malaysia urges fail-safe layers for AI adoption, citing national security risks

Digital Minister Gobind Singh Deo calls for mandatory safety assessments and fail-safe mechanisms before adopting AI, warning of risks to trust and national security.

Industry & market moves

Amazon backs 7.65GW gas plant in Texas for AI data center campus

Amazon is supporting the development of a large off-grid gas plant (GW Ranch) in Pecos County, Texas, to power a new AI data center, reflecting a shift toward dedicated energy solutions for AI workloads.

Omilia raises $67M Series B to scale agentic contact-center AI

Omilia has raised $67 million in Series B funding to expand its self-learning conversational AI for contact centers, aiming for global growth.

OpenAI acquires Next Slide to integrate text-to-slides into ChatGPT Enterprise

OpenAI is acquiring Next Slide and folding its text-to-slide capabilities into ChatGPT Enterprise, aiming to streamline slide creation for business users.

AI product & feature launches

N3XT introduces a Model Context Protocol server that allows AI agents to access live bank data within corporate firewalls, with write capabilities subject to maker-checker approvals.

Firebird launches CIS region's largest AI factory in Armenia

Firebird has launched an AI factory in Armenia, providing large-scale compute infrastructure backed by NVIDIA and Dell, the largest in the CIS region.

Research with immediate practical relevance

AI generates 16 novel viruses from DNA patterns, sparking biosecurity concerns

Researchers used AI trained on 9 trillion nucleotides to create 16 new bacteriophages that could target antibiotic-resistant bacteria, but bioethicists warn guardrails are insufficient.

Explorative modeling emerges as a third axis to boost generative AI performance

A new pre-training technique called explorative modeling adds an exploration phase that probes latent space, improving model quality across modalities by refining representations.