This Week in AI
Big moves in AI agents, coding tools, and enterprise platforms as OpenAI and Anthropic escalate competition, regulators step in, and media, infrastructure, and security grapple with AI’s growing real-world impact.
OpenAI introduces Frontier, an enterprise platform to build/manage AI agents
OpenAI launched Frontier to help enterprises deploy and supervise “AI coworkers” with shared context, onboarding, evaluations/feedback loops, and permission boundaries—positioned as an open platform that can also manage third-party agents.
Read More:
https://openai.com/index/introducing-openai-frontier/
OpenAI ships GPT-5.3-Codex to push “agentic coding” beyond autocomplete
OpenAI released GPT-5.3-Codex as a model tuned for Codex-style agent workflows (planning, mid-turn steering, progress updates), escalating the head-to-head with Anthropic’s latest Claude releases.
Read More:
Anthropic releases Claude Opus 4.6, adding “agent teams” in Claude Code
Anthropic launched Opus 4.6 and highlighted multi-agent “teams” inside Claude Code, plus longer-running task support (including context management/compaction on API) and a detailed safety/system-card style evaluation.
Read More:
https://www.anthropic.com/news/claude-opus-4-6
https://www.anthropic.com/claude-opus-4-6-system-card
Apple’s Xcode 26.3 adds “agentic coding” and support for Codex/Claude via MCP
Apple announced Xcode 26.3 “agentic coding,” letting developers run coding agents (including OpenAI Codex and Anthropic Claude) directly in Xcode, with tool access exposed through the Model Context Protocol (MCP).
Read More:
https://www.apple.com/mz/newsroom/2026/02/xcode-26-point-3-unlocks-the-power-of-agentic-coding/
AI bots become a meaningful share of web traffic, triggering a publisher “defense arms race”
Multiple outlets report a jump in automated AI scraping/agent traffic and more aggressive countermeasures by publishers (rate limits, paywalls, bot detection, legal strategies), citing TollBit’s tracking of bot behavior and bypass tactics.
https://www.wired.com/story/ai-bots-are-now-a-signifigant-source-of-web-traffic/
https://arstechnica.com/ai/2026/02/increase-of-ai-bots-on-the-internet-sparks-arms-race/
“Moltbook,” an AI-only social network, goes viral—and raises safety/authenticity questions
Moltbook (a Reddit-like network for AI agents) drew attention for emergent, meme-like agent behavior—while reporting also questioned how many posts are truly agent-written and highlighted security and misuse risks.
Read More:
https://www.wired.com/story/i-infiltrated-moltbook-ai-only-social-network/,
Bondu AI toy exposure leaks ~50,000 children’s chat logs to anyone with a Gmail account
Reports say Bondu left a web console effectively open, allowing broad access to kids’ conversations and related data; the incident is now drawing political scrutiny and broader calls for stronger security expectations for consumer AI toys.
Read More:
SpaceX–xAI merger fuels “data centers in space” talk (orbital compute + power constraints)
Coverage ties SpaceX’s acquisition of xAI to Musk’s push for space-based compute infrastructure, framed as a response to Earth-side power/cooling constraints for AI—while analysts flag practicality and economics as open questions.
Read more:
