Weekly 3-2-1 AI Brief: 2026-05-09 to 2026-05-16

This Week in AI

This week brought 5 notable AI developments across 3 categories. The highest-scoring signals centred on AI Security.

3 Key AI Developments

1. Google catches first known AI driven zero day discovery in the wild

Google Threat Intelligence Group disclosed the first observed case of an attacker chain using AI to discover and weaponise a zero day. The planned mass attack was halted before execution.

Why it matters: Patch windows already compressed to under 24 hours for known CVEs now compress further when attackers can synthesise novel exploits on demand. Defensive postures predicated on disclosure cycles are obsolete.

2. Anthropic Glasswing gives six enterprise giants early access to Claude Mythos

Anthropic stood up Project Glasswing as a controlled vulnerability discovery programme with six enterprise partners running its unreleased Mythos model.

Why it matters: Frontier model previews are now distributed first to large enterprises with the security capacity to absorb them. Smaller operators wait until the next tier of release.

3. OpenAI B2B Signals: frontier firms use 3.5x more AI per employee than typical

OpenAI quantified the AI per employee gap between frontier and typical firms at 3.5x, based on aggregate token usage across its enterprise customer base.

Why it matters: The intensity gap is now measurable. Boards and investors will use this ratio as a leading indicator the way revenue per employee already gets used.

2 Interesting Pieces

SailPoint launches Agentic Fabric to secure non human AI agent identities at enterprise scale

Source: SailPoint / Yahoo Finance

SailPoint launched Agentic Fabric, an identity centric solution for securing AI agents and other non human identities. It gives organisations the context to understand what agents can access, who is responsible for them, and how to govern them at scale.

OpenAI launches Development Company with 19 partners and acquires consultancy Tomoro

Source: OpenAI / CNBC

OpenAI announced a new business unit including the acquisition of applied AI consulting firm Tomoro. The OpenAI Development Company is a partnership with 19 investment and consultancy firms including Bain, Goldman Sachs, and SoftBank, majority owned and controlled by OpenAI.

1 Actionable Idea

Signal Summary

| Signal | Category | Company | Score | |--------|----------|---------|-------| | Google catches first known AI driven zero day discovery in the wild | AI Security | Google | 8.4 | | Anthropic Glasswing gives six enterprise giants early access to Claude Mythos | Model Releases | Anthropic | 7.4 | | OpenAI B2B Signals: frontier firms use 3.5x more AI per employee than typical | Enterprise AI | OpenAI | 7.4 | | SailPoint launches Agentic Fabric to secure non human AI agent identities at enterprise scale | AI Security | SailPoint | 7.3 | | OpenAI launches Development Company with 19 partners and acquires consultancy Tomoro | Enterprise AI | OpenAI | 7.1 |

Citable Summary

Week: 2026-05-09 to 2026-05-16

Signals included: 5

Average composite score: 7.51

Categories covered: AI Security, Model Releases, Enterprise AI

Source: David and Goliath AI Intelligence Engine