AI Daily Dev — Week 20

DeepMind's AI Co-Mathematician Sets a New High on FrontierMath Tier 4

Multi-agent workbench built on Gemini 3.1 — a project coordinator fans research questions out to sub-agents that write code, search literature, and attempt proofs.
Solved 23 of 48 problems (48%) on Epoch AI's FrontierMath Tier 4, more than double the 19% scored by the underlying Gemini 3.1 Pro and a new SOTA.
Oxford's Marc Lackenby cracked Problem 21.10 from the Kourovka Notebook after a reviewer agent flagged a flaw in the AI's first proof attempt and surfaced a strategy he could finish.
Stateful, asynchronous workspace that tracks failed hypotheses and outputs LaTeX with margin annotations — DeepMind frames it as 'Claude Code for mathematicians.'

research arxiv.org

Google Threat Intelligence Group says it disrupted the first AI-built zero-day used in a planned mass-exploitation campaign.
The exploit, written in Python with help from cybercrime model OpenClaw, bypasses 2FA on a widely deployed open-source web admin tool.
Attackers tipped Google off by leaving a hallucinated CVSS score inside the script; Google notified the vendor before any victims were hit.
GTIG analyst John Hultquist: "It's here. The era of AI-driven vulnerability and exploitation is already here."
Front-page coverage at CNBC, Bloomberg, NBC News, and Bleeping Computer; on HN with active debate about Anthropic's Mythos and Project Glasswing.

industry cloud.google.com

Claude Platform on AWS is now GA — Anthropic's full native platform (Messages API, Console, Managed Agents, Agent Skills, MCP connector, web search/fetch, code execution) accessed through an AWS account.
Billing, IAM auth, and CloudTrail logging route through AWS; Anthropic still operates the service and processes data outside the AWS security boundary.
Opus 4.7, Sonnet 4.6, and Haiku 4.5 available at launch across 17 AWS regions; new models ship there as they release.
AWS is the first cloud provider to host the native Claude Platform, distinct from Bedrock; follows Anthropic's $100B+ AWS compute commitment.
On HN front page (item 48103042) with active back-and-forth about what "on AWS" actually means.

industry anthropic.com

Android Show I/O Edition reveal: Googlebook laptops, an Android 17-based desktop OS (codename Aluminium), and Gemini Intelligence across phones, Wear OS, Auto, and glasses.
First Googlebooks ship this fall with Acer, ASUS, Dell, HP, and Lenovo; each has a 'glowbar' on the lid and runs Android apps natively, with files synced from Android phones.
Create My Widget lets users describe a widget in natural language (TechRadar calls it the 'first step in generative UI'); Gemini Intelligence rolls to Galaxy S26 and Pixel 10 this summer.
On HN front page (item 48111545): 749 points, 464 comments.

industry blog.google

New packaged install puts Claude inside QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365.
15 ready-to-run agentic workflows across finance, ops, sales, marketing, HR, and customer service — payroll planning, invoice chasing, month-end close, campaign management.
Runs inside Claude Cowork; user approves each plan before anything sends, posts, or pays; account permissions carry over from the connected tool.
No surcharge on top of existing Claude licenses; a 10-city free fluency-training tour kicks off today (May 14) in Chicago.

tools anthropic.com

Four-year deal: grant dollars, Claude usage credits, and Anthropic engineering time pooled with the Gates Foundation.
Health gets the largest slice — drug discovery for polio, HPV, and eclampsia/preeclampsia, and tooling so ministries in low- and middle-income countries can act on health data.
Funds open datasets and benchmarks for dozens of underserved African languages so any lab can train better translation and tutoring models.
K–12 tutoring and college-advising tools targeted at the US, sub-Saharan Africa, and India; HN thread on the announcement is debating how much of the headline figure is real cash vs. credit retirement.

industry anthropic.com

New cybersecurity platform that builds an editable threat model from your repo, validates exploitable vulnerabilities in isolated sandboxes, and drafts patches with audit-ready evidence.
Built on three models: stock GPT-5.5, GPT-5.5 with Trusted Access for Cyber for verified defenders, and GPT-5.5-Cyber — a permissive variant for red-team and pentest work.
Launch partners include Cloudflare, Cisco, CrowdStrike, Palo Alto Networks, Oracle, Akamai, Fortinet, Zscaler, Snyk, Semgrep, and Socket.
Reads as OpenAI's direct response to Anthropic's Claude Mythos and Project Glasswing; available by request, broader rollout planned with industry and government partners.

tools openai.com

Akamai's largest contract ever — roughly 6% of its annual revenue once fully ramped — to host Claude inference on Akamai's edge cloud.
Anthropic CEO Dario Amodei cited 80x year-on-year growth in annualized revenue and usage in Q1; first economic impact expected by end of 2026.
Adds to Anthropic's compute scramble alongside SpaceX Colossus 1 and the $100B+ AWS commitment.
Akamai stock closed Friday up 27% at $148.38 on the news.

industry bloomberg.com

Open-OSS/privacy-filter typosquatted OpenAI's real release, copied the model card verbatim, and shipped a loader.py that pulled a Rust info-stealer for Windows.
Payload disables AMSI and ETW, runs VM and debugger checks, and runs eight parallel collectors targeting Chromium, Firefox, Discord, and crypto wallets.
HiddenLayer telemetry linked six more malicious repos under the same account, all uploaded April 24 — likely a coordinated supply-chain campaign.
Spent days on Hugging Face Trending before removal; covered by BleepingComputer, The Hacker News, and Security Boulevard.

community hiddenlayer.com

New OpenAI-controlled entity ("DeployCo") launches with $4B+ backing to embed Forward Deployed Engineers inside enterprises and rebuild workflows around GPT-5.5.
Founding investors led by TPG with Advent, Bain Capital, Brookfield, plus Capgemini, Bain & Company, and McKinsey as integration partners; $10B pre-money.
Acquires London-based applied-AI firm Tomoro, adding ~150 FDEs and Deployment Specialists with prior work for Tesco, Virgin Atlantic, and Supercell.
Mirrors Palantir's FDE playbook; on HN (item 48094531).

industry openai.com