March 1, 2026 • February Recap
Top stories & highlights — February 2026
Below are 18 notable stories across Medicine, Biology, AI, Science, and Tech that shaped February 2026. Each item is a short summary with links where available.
- Outlook add-in hijack used in phishing Attackers abused a legitimate Outlook add-in to create convincing phishing flows — a reminder to audit add-in permissions and enforce conditional access.
- Google Chrome zero-day (CVE-2026-2441) A critical CSS rendering vulnerability saw active exploitation; update browsers and embedded WebViews promptly.
- Password-manager recovery attacks Researchers disclosed attacks targeting password-recovery flows in cloud managers (Bitwarden, Dashlane, LastPass). Strengthen account recovery and MFA practices.
- OpenAI's "Lockdown Mode" & elevated-risk labels OpenAI introduced new safety controls to mitigate advanced prompt-injection and risky content generation in enterprise contexts.
- AI recommendation poisoning Platforms found examples of businesses and actors manipulating summarization and recommendation prompts; defensive prompt design and provenance checks are becoming standard.
- Infosys partners with Anthropic A major enterprise partnership to build agent-based AI solutions and accelerate production use-cases.
- Cohere's open multilingual models New multilingual checkpoints make localization easier for chat and retrieval tasks.
- Ricursive Intelligence funding round Another large AI funding milestone (hundreds of millions) highlighted continued investor appetite for specialized model plays and infrastructure.
- Adani's $100B AI data center pledge Large-scale infrastructure investments signaled national-level plays for AI compute capacity.
- Heretic & open-model tooling trends Open-source projects addressing censorship, tooling, and agent orchestration climbed trending lists; community-driven tooling remains vibrant.
- Fast-growing MCP tooling MCP-related SDKs and tools (Python/TypeScript SDKs, fastmcp) expanded, making it practical to build context-aware agent systems.
- OpenClaw/infostealer campaigns Threat actors shifted focus from browser creds to AI agent artifacts and tokens, increasing the need for secrets hygiene and local sandboxing.
- Apple tests E2E RCS messaging Privacy-forward messaging experiments showed industry-level interest in secure carrier messaging.
- Mixtral / Mistral model releases New performant open models and instruction-tuned variants improved local deployment options for developers.
- Superpowers agentic skills framework Agent skill frameworks made re-usable behaviors and tool integrations simpler across agents and projects.
- CRISPR delivery advances Preclinical results suggested improved delivery vectors that reduce off-target edits — a step forward for safer gene therapies.
- mRNA therapeutics beyond vaccines mRNA platforms expanded into rapid-response therapeutics and oncology candidates, with multiple phase I trials announced.
- Diagnostics with AI AI-assisted imaging tools showed higher detection rates for rare conditions in large benchmark studies; adoption depends on regulatory pathways.
Key takeaways for projects
- Prioritize security for integrations (add-ins, OAuth, recovery flows) and rotate credentials regularly.
- Adopt defensive prompt patterns and provenance tracking for model-driven features.
- Leverage open-source MCP tooling and local model-serving options for reproducible deployments.
- Track biotech developments (CRISPR delivery, mRNA therapeutics) for content and tutorial opportunities.
- Keep `AI Tools & Resources` updated with fresh links and highlight new additions with a "new" tag.
Sources: industry roundups from February 2026, vendor blogs, and open-source project feeds.