Five stories for studios and agent builders, from engine code written by models to storefront politics and a multi-agent hacking campaign.
1. Epic confirms AI is now writing Unreal Engine code
Epic Games vice president of development Marcus Wassmer confirmed that the company is "definitely using AI to develop code and features for Unreal Engine," and reports on Epic's gen-AI push add that its engineers "are no longer writing the bulk of the code." Epic also credits AI tools with squashing bugs that had lived in the engine for a decade, and says the same approach will extend across Fortnite and Unreal Engine 6.
Why it matters: When the maker of one of the world's biggest game engines trusts agents with core engine code, agentic development stops being an experiment and becomes the baseline studios are measured against.
2. Valve is on Steam's visibility problem but won't gatekeep AI games
Valve says "a significant number" of its staff are working to fix Steam's new game visibility dilemma, after developers complained that events like Steam Next Fest were crowded with AI slop. The company stated it does not plan to start "gatekeeping" games made with AI. A follow-up report adds that indie developers are now receiving threats that their games will be cloned unless they pay a ransom.
Why it matters: Discoverability is the next battleground for small studios. With Valve declining to filter AI-made titles, visibility strategy on storefronts matters as much as build quality.
3. A leaked prompt shows Meta's shopping agent playing favorites
WIRED reports that a leaked set of instructions tells Meta's AI agent to fill its top picks with well-known stores and to check only three websites per request, just as millions of shoppers hand the agent their holiday lists. The document suggests the agent's "choices" are heavily scripted rather than neutral comparisons.
Why it matters: Agents are becoming a distribution channel with their own politics. If your store, game, or service is not in an agent's whitelist, customers may never see it, which is why agent-facing optimization deserves early attention.
4. A hacking campaign ran on 44 coordinated AI agents
The Hacker News details CyberStrikeAI, an operation that orchestrates 44 AI agents across the full attack lifecycle: reconnaissance, vulnerability discovery and validation, exploitation, command and control, persistence, data exfiltration, and automated reporting, including fake Japanese language login pages aimed at targets in Japan.
Why it matters: The same multi-agent orchestration that powers game pipelines works for attackers too. Studios and agent builders running agent fleets should assume adversaries are doing the same and harden keys, credentials, and CI access accordingly.
5. Prompt-to-game platforms want to host the agent workforce
Naavik surveys the prompt-to-game landscape and spotlights Playgama's new toolset that lets AI agents publish, analyze, and iterate on HTML5 games without human intervention, including an MCP server that pushes builds straight from Claude Code or Codex and returns a live shareable link in minutes.
Why it matters: The bottleneck for agent-built games is shifting from creation to distribution. Platforms exposing agent-friendly publishing APIs are positioned to capture the volume wave coming out of coding agents.
AgentLayer tracks this shift daily: join the portal where studios and agent builders build, play, and monetize together.