Eyes on the Chaos
Sunday, May 17, 2026

Archived edition

Sunday, May 17, 2026

8 stories curated from 16 sources

In today's issue

DesignEthicsProduct
  1. 01
    Research repository ArXiv will ban authors for a year if they let AI do all the work

    ArXiv cracks down on careless LLM use in scientific papers with year-long bans.

  2. 02
    OpenAI co-founder Greg Brockman takes charge of product strategy

    Brockman's new role comes as OpenAI plans to merge ChatGPT with Codex.

  3. 03
    The undo problem in AI products

    AI-generated content creates complex challenges for traditional undo functionality in products.

  4. 04
    html.to.design — New Chrome extension feature

    Chrome extension captures webpages and pastes them directly into Figma for editing.

  5. 05
    How to write a DESIGN.md file Claude can actually use

    Guide for creating design documentation that AI tools can effectively interpret and use.

  6. 06
    How the pursuit of optimisation reshaped art, aesthetics and us

    Artists and designers question what gets lost as frictionlessness colonizes everyday experiences.

  7. 07
    Did You Mean to Share That Selfie? What to Know About Instagram Instants.

    Instagram's new feature instantly blasts photos to followers, raising privacy concerns.

  8. 08
    A catechism for robots

    Autonomous systems need core moral values beyond just logical rules to function properly.

AI Research & News

Research repository ArXiv will ban authors for a year if they let AI do all the work

TechCrunch

Ethics

ArXiv cracks down on careless LLM use in scientific papers with year-long bans.

  • New enforcement: The preprint server is taking stricter action against authors who rely entirely on AI to write research papers.
  • Year-long penalty: Authors caught letting AI do all the work face a one-year ban from submitting to the platform.
  • Quality concern: The move addresses growing concerns about AI-generated content diluting scientific research quality.

For ethics

Consider similar content quality standards for your team's AI-assisted documentation and research outputs.

OpenAI co-founder Greg Brockman takes charge of product strategy

TechCrunch

Product

Brockman's new role comes as OpenAI plans to merge ChatGPT with Codex.

  • Leadership shift: Co-founder Greg Brockman is now leading product strategy in OpenAI's latest organizational change.
  • Product consolidation: The company reportedly plans to combine ChatGPT with its programming tool Codex into a unified offering.
  • Strategic focus: The move signals OpenAI's push toward more integrated product experiences rather than standalone tools.

For product

Watch for unified AI tooling trends that could affect your team's workflow consolidation decisions.

Product & UX

The undo problem in AI products

UX Collective

DesignProduct

AI-generated content creates complex challenges for traditional undo functionality in products.

  • Traditional undo breaks: Standard undo/redo patterns don't work well when AI generates unpredictable or large amounts of content.
  • User expectations: People expect to reverse AI actions, but the non-deterministic nature makes exact reversal impossible.
  • Design challenge: Teams need new interaction patterns that account for AI's probabilistic outputs and iterative workflows.

For design

Audit your AI features now for undo patterns—this will become a bigger user frustration as AI adoption grows.

html.to.design — New Chrome extension feature

Sidebar.io

Design

Chrome extension captures webpages and pastes them directly into Figma for editing.

  • Workflow shortcut: The extension lets designers capture entire webpages or sections with one click.
  • Figma integration: Captured content goes straight to clipboard and pastes as editable elements in Figma.
  • Prototyping speed: Eliminates the screenshot-trace-recreate cycle when working with existing web interfaces.

For design

Could significantly speed up competitive analysis and design system audits—worth piloting with your design teams.

How to write a DESIGN.md file Claude can actually use

UX Collective

DesignProduct

Guide for creating design documentation that AI tools can effectively interpret and use.

  • AI-readable docs: Structure design documentation so AI assistants like Claude can understand and work with your product decisions.
  • Documentation format: Specific guidelines on how to organize design rationale, constraints, and system information for AI consumption.
  • Cross-functional value: Well-structured DESIGN.md files help both human teams and AI tools understand product context.
How the pursuit of optimisation reshaped art, aesthetics and us

Sidebar.io

Design

Artists and designers question what gets lost as frictionlessness colonizes everyday experiences.

  • Frictionless everything: The drive to optimize and remove friction is reshaping how we interact with art, products, and each other.
  • Cultural pushback: Artists, designers and theorists are questioning whether seamless experiences eliminate meaningful human elements.
  • Design philosophy: The piece challenges the assumption that removing all friction creates better experiences.

For design

Consider where intentional friction might improve user experience—not every interaction should be optimized away.

Business & Strategy

Did You Mean to Share That Selfie? What to Know About Instagram Instants.

NYT Technology

ProductEthics

Instagram's new feature instantly blasts photos to followers, raising privacy concerns.

  • Instant sharing: The feature automatically sends photos to mutual followers or close friends without traditional posting friction.
  • Privacy risk: Users may accidentally share content they didn't intend to distribute widely due to the immediate nature.
  • Design tension: Instagram prioritizes engagement speed over user control and reflection time.

For product

Review your own instant-share features for unintended privacy consequences—the engagement boost may not be worth user trust issues.

A catechism for robots

Sidebar.io

EthicsProduct

Autonomous systems need core moral values beyond just logical rules to function properly.

  • Beyond logic: Autonomous AI systems require fundamental moral principles, not just programmed rules and constraints.
  • Value alignment: The challenge is encoding human values that can guide decisions in unexpected situations.
  • System design: Companies building AI products need frameworks for embedding ethical decision-making into autonomous features.

For product

Start documenting your team's core values for AI features now—you'll need them as your products become more autonomous.