AI/Tech Brief — 2026-05-26

1. TL;DR

• Pre-installed apps on high-end Motorola devices are intercepting Amazon app launches to inject affiliate codes.
• California amends its Digital Age Assurance Act to protect open-source operating systems like Linux from impossible compliance burdens.
• A critical flaw in AWS HTTP API Gateway allowed attackers to bypass JWT auth simply by adding a trailing slash.

2. Key stories

• Motorola Hijacks Amazon App for Affiliate Revenue: Recent updates to Motorola’s pre-installed “Smart Feed” app intercept Amazon app launches to inject affiliate tracking codes. Why it matters: Highlights aggressive and invasive monetization tactics employed by OEMs, even on premium $1,900 devices. Source
• California Exempts Linux from Age Verification: Following community backlash, lawmakers introduced AB 1856 to exempt open-source operating systems from the upcoming Digital Age Assurance Act. Why it matters: Protects the FOSS ecosystem from impossible compliance requirements while leaving proprietary stores accountable. Source
• AWS API Gateway Auth Bypass: A security researcher earned a $12K bounty by bypassing JWT authentication in AWS HTTP API Gateway by appending a trailing slash to a route. Why it matters: Exposes the hidden risks of greedy path matching and path rewriting in managed cloud infrastructure. Source
• CockroachDB’s Scalable Vector Indexing: ByteByteGo published a deep dive into how CockroachDB engineered its vector indexing to support AI workloads at scale. Why it matters: Offers practical insights into database performance optimization for modern LLM infrastructure and AI applications. Source
• The AI Bubble vs. The Dot-Com Bubble: Cory Doctorow contrasts the early web with the current AI boom, arguing that AI adoption is top-down and focused on throughput rather than quality. Why it matters: Provides a skeptical counter-narrative to the AI hype cycle, highlighting the poor unit economics of current generative models. Source

3. Quiet but interesting

• Using AI to Write Better Code More Slowly: Nolan Lawson argues against using AI as a “slop cannon” for raw throughput, instead advocating for multi-model PR reviews and deeper codebase understanding. Why it matters: Signals a shift among experienced engineers from using AI for speed to using it for code quality and review. Source
• Norway Adopts 2PB of Huawei Storage for LLMs: A Norwegian research cluster opted for Huawei flash storage to support their AI training infrastructure. Why it matters: Shows that performance and cost can sometimes outweigh geopolitical pressures in large-scale AI hardware procurement. Source

4. Skip

• Major AI Labs Go Quiet: A quiet 24 hours for official product releases and changelogs. OpenAI, Anthropic, Google DeepMind, and Gemini CLI all had zero updates today.
• Mullvad VPN Exit IP Mitigation: Routine rollout of mitigations for exit IP VPN servers; relevant only to specialized infosec and networking administrators.