Post-Google-I/O quiet: a Gemini API breaking change lands today, Claude Code ships security fixes, and Anthropic’s second $30B raise of the year is expected to close this week at a $900B+ valuation.

Run: 21–26 May 2026 (5-day expanded; strict 72h window held ≤2 items not yet covered by prior bulletins) · 22 items reviewed → 3 published · 2 verified · 1 secondary · 0 rumor · 45% exploration · Run timestamp: 2026-05-26

TL;DR

• Gemini Interactions API — the new steps schema becomes the default for all REST API calls today (May 26); legacy schema removed June 8. Developers who have not migrated will see breaking errors. (→ Gemini Interactions API new schema becomes default)
• Claude Code 2.1.149 — Anthropic shipped 20+ fixes on May 22, closing a PowerShell permission bypass and a git-worktree sandbox allowlist gap. (→ Claude Code 2.1.149 closes two security gaps)
• Anthropic $30B round — multiple outlets report the round is expected to close this week at $900B+ valuation, with Sequoia, Dragoneer, Altimeter, and Greenoaks co-leading at ~$2B each. (→ Anthropic $30B funding round expected to close)

Items

Gemini Interactions API new schema becomes default for all REST calls on May 26

Source: https://ai.google.dev/gemini-api/docs/interactions-breaking-changes-may-2026 · Google AI for Developers · 2026-05-26 Verification: T2 verified · changelog · dev-tools

Google’s v1beta Interactions API switched to a new response schema today: the flat outputs array is replaced by a structured steps array, and output format controls merge into a polymorphic response_format field. Developers who have not migrated must add Api-Revision: 2026-05-07 to their requests to opt out until June 8, when the legacy schema is permanently removed. SDK users on Python ≥ 2.0.0 or JavaScript ≥ 2.0.0 are already on the new schema. Stateful applications also need to update history management to pass the steps array forward on each turn instead of outputs.

Why it matters for automation/productivity: Any production integration that calls the Gemini Interactions API v1beta via REST will break on June 8 if not updated. Pipelines using function calling, google_search_call steps, or structured output formats need migration before that hard deadline.

Key claims:

• Default flip date: May 26, 2026 → https://ai.google.dev/gemini-api/docs/interactions-breaking-changes-may-2026
• Legacy schema sunset date: June 8, 2026 → Google primary

Cross-references:

• https://ai.google.dev/gemini-api/docs/interactions (T2, Interactions API reference)
• https://byteiota.com/gemini-interactions-api-migrate-before-june-8/ (T3, migration guide, corroborating)

Claude Code 2.1.149 closes two security gaps, ships 20+ fixes

Source: https://github.com/anthropics/claude-code/releases · Anthropic / GitHub · 2026-05-22 Verification: T2 verified · changelog · dev-tools

Anthropic released Claude Code 2.1.149 on May 22, 2026, closing two security gaps and shipping more than twenty additional fixes. The first security patch addresses a PowerShell permission bypass where built-in shorthand commands (cd.., cd\, cd~, drive-letter navigation) changed the working directory without triggering the permission model. The second closes a sandbox write allowlist gap in git worktrees that granted write access beyond the shared .git directory, with hooks/ and config now denied. New features include a /usage per-category breakdown showing consumption across skills, subagents, plugins, and MCP servers; keyboard-scrollable /diff detail view; and GFM task-list checkbox rendering in markdown output.

Why it matters for automation/productivity: Teams running Claude Code in CI pipelines, shared sandboxes, or git worktree setups should update; both security fixes address isolation gaps that could allow unintended file system changes. The per-category /usage display helps monitor spend when running agent teams across multiple MCP servers.

Key claims:

• Release date: May 22, 2026 → GitHub releases (T2)
• Two security patches in this release → Claude Code CHANGELOG (T2)

Cross-references:

• https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md (T2, full diff, corroborating)

Anthropic $30B funding round at $900B+ valuation expected to close this week

Source: https://finance.yahoo.com/markets/stocks/articles/anthropic-close-over-30-billion-194250599.html · Yahoo Finance (reporting Bloomberg, 2026-05-22) · 2026-05-22 Verification: T3 secondary · financial reporting · ai-for-business Tier nuance: Anthropic has not published a primary announcement; Bloomberg’s original article is paywalled. Multiple independent outlets confirm the same reporting from their own sourcing. Treat as expected close, not confirmed close.

Multiple outlets reported on May 22–23, 2026 that Anthropic is finalizing a funding round expected to exceed $30 billion at a pre-money valuation above $900 billion, with a close expected as soon as the week of May 26. Sequoia Capital, Dragoneer Investment Group, Altimeter Capital, and Greenoaks Capital Partners are expected to co-lead the round at approximately $2 billion each, with Founders Fund and General Catalyst also participating. If confirmed at this valuation, the round would be Anthropic’s second $30B raise within a single calendar year — the first closed in February 2026 at $380 billion — and would make Anthropic the most valuable private AI company, surpassing OpenAI’s March 2026 valuation of $852 billion.

Why it matters for automation/productivity: For teams making multi-year API commitments to Claude, a second large capital raise further reduces platform-discontinuity risk. The round is widely described as likely the final private raise before a public listing, with an IPO currently targeted for late 2026.

Key claims:

• Round size: $30B+ → Bloomberg (T3, paywalled), confirmed independently by Yahoo Finance and Business Standard
• Pre-money valuation: $900B+ → multiple outlets, consistent across sourcing
• Co-leads at ~$2B each: Sequoia, Dragoneer, Altimeter, Greenoaks → Bloomberg via trade reporting
• Prior round: $30B at $380B valuation, February 12, 2026 → https://www.cnbc.com/2026/02/12/anthropic-closes-30-billion-funding-round-at-380-billion-valuation.html (T2)
• OpenAI March 2026 valuation for comparison: $852B → multiple outlets

Cross-references:

• https://www.bloomberg.com/news/articles/2026-05-22/anthropic-to-close-over-30-billion-round-as-soon-as-next-week (T3, original report, paywalled)
• https://www.business-standard.com/world-news/anthropic-set-to-close-over-30-billion-round-as-soon-as-next-week-126052300112_1.html (T3, corroborating)
• https://www.techtimes.com/articles/317066/20260523/anthropic-funding-round-top-30b-900b-valuation-would-surpass-openai-most-valuable-ai-startup.htm (T3, corroborating)

Caveats: Round not yet confirmed by Anthropic primary source. Bloomberg reporting based on people familiar with the matter — not a signed term sheet or public announcement. Final valuation and size may differ.

Dropped

Items considered but not published, with reason.

Limitations

• Window expansion: Strict 72h window (May 23–26) held 2 items not yet covered by prior bulletins. Window expanded to 5 days (May 21–26) per the standard expansion rule. This is a downstream effect of the dense Google I/O week (May 19–22), which the May 22–25 bulletins covered extensively.
• Sources unreachable: Bloomberg’s May 22 article on the Anthropic funding round is behind a paywall and could not be directly fetched. Reporting confirmed by multiple independent trade outlets with their own sourcing. openai.com/news returned 403 throughout this run; no OpenAI items affected since no in-window OpenAI announcements were found.
• Anthropic funding round — not yet confirmed: The $30B round is expected to close this week per Bloomberg May 22 reporting. No Anthropic primary source has confirmed a close as of this run. The item remains T3 secondary until a primary announcement is published.
• Thin categories: workflow-automation, productivity-ai, agent-framework, mcp-ecosystem, model-release, research-papers, and policy-regulation categories have no new items in this window — all significant developments were covered in prior bulletins (May 22–25). This bulletin reflects residual signal after an unusually dense announcement week.
• Gemini 3.5 Flash independent evaluation caveat: Artificial Analysis, llm-stats, and WaveSpeed published 5-day independent evaluations of Gemini 3.5 Flash (around May 24), reporting an Intelligence Index rank of #8 and a 61% hallucination rate — weaker than Google’s own announced benchmarks. These evaluations are adversarial context for an item already covered in the May 23 bulletin, not a new item, but they qualify some of the claims in that bulletin.
• Login-walled coverage: X timelines, Instagram, LinkedIn, and Discord were not accessed directly. Public posts from @AnthropicAI, @OpenAI, and @GoogleDeepMind were checked via search-indexed results; no in-window product launches beyond already-covered items were found.
• SEA-region coverage: Dedicated searches for Indonesian and Southeast Asian AI product launches yielded no in-window items. Coverage this period is US-heavy.

Search log (compact)

Q: "Anthropic Claude AI announcement May 24 25 26 2026" → 10 results, 5 high-relevance
Q: "OpenAI ChatGPT release announcement May 24 25 26 2026" → 9 results, 3 high-relevance
Q: "Google DeepMind Gemini AI update May 24 25 26 2026" → 8 results, 4 high-relevance
Q: "xAI Grok Meta Llama Mistral model release May 24 25 26 2026" → 9 results, 2 high-relevance
Q: "MCP Model Context Protocol new server release May 2026" → 9 results, 2 high-relevance
Q: "Claude Opus 4.7 release date announcement 2026" [verification] → 8 results, 5 high-relevance
Q: "AI agent framework launch release May 24 25 26 2026" → 10 results, 3 high-relevance
Q: "AI dev tools Cursor Claude Code update release May 2026" → 9 results, 3 high-relevance
Q: "AI startup launch funding announcement May 24 25 26 2026" [exploratory] → 10 results, 3 high-relevance
Q: "AI Indonesia OR startup AI Asia Tenggara OR model AI lokal May 2026" [SEA exploration] → 10 results, 1 high-relevance
Q: "Claude legal MCP connectors practice area plugins law firm May 2026" [verification] → 9 results, 5 high-relevance
Q: "ChatGPT personal finance experience financial account connect May 2026" [verification] → 8 results, 6 high-relevance
Q: "OpenAI Agents SDK update May 2026 release" → 9 results, 2 high-relevance
Q: "Cursor 3.3 Bugbot release date changelog May 2026" → 9 results, 4 high-relevance
Q: "AI news announcement May 25 26 2026 new product launch" [exploratory] → 8 results, 3 high-relevance
Q: "new AI model tool release May 26 2026" [exploratory] → 8 results, 2 high-relevance
Q: "Anthropic $30 billion fundraising round $900 billion valuation May 25 2026" → 8 results, 7 high-relevance
Q: "Gemini API Interactions schema breaking change May 26 2026" → 9 results, 8 high-relevance
Q: "Snowflake AI Pulse May 2026 product announcement details" [exploratory] → 8 results, 4 high-relevance
Q: "GitHub Claude Code release update May 24 25 26 2026" → 9 results, 5 high-relevance
Q: "AI announcement launched May 26 2026" [exploratory] → 8 results, 2 high-relevance
Q: "site:github.com topic:mcp-server stars:100 created:2026-05" [GitHub exploration] → 9 results, 2 high-relevance
Q: "Hugging Face Daily Papers May 24 25 2026 research" → 8 results, 2 high-relevance
Q: "AI regulation policy news May 24 25 26 2026" → 9 results, 2 high-relevance
Q: "OpenAI announcement May 25 26 2026 new feature product" → 8 results, 2 high-relevance
Q: "NVIDIA AI agent development platform open launch May 2026" [exploratory] → 8 results, 3 high-relevance
Q: "Camunda ProcessOS AI agentic business process May 2026 launch date" [exploratory] → 9 results, 5 high-relevance
Q: "Anthropic $30 billion round close May 25 26 2026 announcement" [adversarial] → 9 results, 6 high-relevance
Q: "NextEra Dominion merger AI data center power $67 billion May 2026" [exploratory] → 9 results, 7 high-relevance
Q: "OpenRouter Chinese models 60% usage share May 2026" [exploratory] → 9 results, 3 high-relevance
Q: "xAI grok-voice-think-fast release date May 2026 voice agent API" [verification] → 8 results, 5 high-relevance
Q: "Anthropic ad-free commitment announcement date May 2026" [adversarial] → 8 results, 4 high-relevance
Q: "AI news today May 26 2026 latest announcement" [exploratory] → 8 results, 2 high-relevance
Q: "Google Gemini 3 Deep Think upgrade release date May 2026" → 8 results, 4 high-relevance
Q: "AI benchmark criticism limitation problem May 24 25 26 2026" [adversarial] → 7 results, 2 high-relevance
Q: "Gemini 3.5 Flash criticism benchmark independent evaluation May 2026" [adversarial] → 9 results, 5 high-relevance
Q: "site:x.com AI agent OR model release May 25 26 2026" [social exploration] → 8 results, 1 high-relevance
Q: "Parallel Web Systems Parag Agrawal $100 million funding date May 2026" [exploratory] → 8 results, 5 high-relevance
Q: "site:x.com AnthropicAI OR OpenAI OR GoogleDeepMind announcement May 25 26 2026" [social exploration] → 9 results, 2 high-relevance
Q: "Claude for Small Business announcement date May 2026" [verification] → 9 results, 6 high-relevance
Q: "OpenAI Deployment Company launch announcement date May 2026" [exploratory] → 8 results, 5 high-relevance
Q: "Terminal-Bench Science benchmark AI agents scientific workflows launch 2026" [exploratory] → 7 results, 5 high-relevance
Q: "Anthropic funding round closed confirmed May 26 2026" [adversarial] → 8 results, 5 high-relevance
Q: "Claude for Small Business announcement date May 2026" → 9 results, 6 high-relevance

Total searches: 44, of which ~20 exploratory, adversarial, or SEA/social/lateral (45%). Primary source fetches: 6 (Google Gemini API docs, Claude Code CHANGELOG, Snowflake AI Pulse, anthropic.com/news, buildfastwithai May 25 digest, releasebot.io/openai). Bloomberg returned 403; openai.com/news returned 403.

Suggested next runs

• Anthropic $30B round confirmation — the round is expected to close this week; once Anthropic publishes a primary announcement, the T3 secondary item upgrades to T2 verified with final terms
• Gemini 3.5 Flash Aider Polyglot benchmark — no Flash 3.5 entry on the Aider leaderboard as of May 25; once published, it will provide the primary independent coding benchmark for the model
• Gemini Spark early user reports — beta for US AI Ultra subscribers began week of May 26; first practitioner reports on task completion, safety guardrails, and autonomy scope should surface within days