AI News: June 21, 2026

1. NYU’s Damodaran warns an AI crash could hit harder than the dot-com bust

Industry. NYU finance professor Aswath Damodaran argues that the heavy debt financing behind today’s AI infrastructure buildout makes a potential AI crash more dangerous than the equity-driven dot-com bust. He singles out AI-driven job displacement as a structural risk to the business models funding the boom. The caution is a useful counterweight for practitioners weighing the durability of current AI capital spending. Source

2. OpenAI tripled Q1 revenue to $5.7B while burning $3.7B

OpenAI. OpenAI’s first-quarter 2026 revenue tripled year over year to $5.7 billion against $3.7 billion in expenses, including $2.3 billion in stock-based compensation, with roughly $73 billion in reserves remaining. The figures, reported by The Decoder, illustrate both the rapid commercial growth and the heavy cash burn underpinning frontier-model economics. Source

3. Data2Story turns a CSV into a verified interactive news article with seven AI agents

Research. Researchers from Oxford and Stanford introduced Data2Story, a seven-agent pipeline that converts a raw CSV into a data-journalism article with sources verified for 93% of statements, reportedly matching human quality on complex reports. It is a concrete demonstration of multi-agent systems applied to fact-grounded, verifiable content generation. Source

4. EU AI Act’s unclear “deepfake” definition becomes a problem for retail

Regulation. Eurocommerce is seeking exemptions from EU AI Act transparency rules for AI-generated marketing content, arguing that product imagery does not constitute a deepfake, with Zalando noting that roughly 90% of its marketing already uses AI generation. The dispute highlights real ambiguity in AI-labeling regulation that will shape how companies disclose AI-generated media. Source