Court filings unsealed in a California case reveal Meta employees in 2023 discussed a 'Whisper strategy' aimed at discrediting Jonathan Haidt, the author whose research links social media to teen mental health problems. One idea floated was recruiting an academic, possibly Columbia's Andrew Gelman, to write a critical takedown of Haidt's work. It's unclear whether Meta ever carried out the plan, and the company declined to comment.
theverge.com
· 2026-11-13
Court documents reportedly reveal Meta employees discussed a 'whisper campaign' approach aimed at undermining social psychologist Jonathan Haidt, a prominent critic of social media's effects on youth mental health. The filings suggest internal communications explored ways to counter Haidt's influence rather than directly engage his research publicly.
tech.slashdot.org
· 2026-11-01
A developer released Overtone, an open-source captioning system that stores transcribed words with timing and speaker data once, then generates separate caption cuts optimized for different screen shapes like 16:9 and 9:16 from a single file. It ships as a CLI for creating or re-cutting caption files from subtitles, audio, or video, plus a video.js player plugin that displays the caption version matching the player's shape. The tool supports offline re-cutting of existing subtitles, and can transcribe audio using OpenAI or Azure APIs, or run fully offline via whisper.cpp and Ollama.
github.com
· 2026-10-29
Court filings in litigation over Meta's teen safety claims reveal internal 'Well-being Meeting' notes from March 2023 describing a 'Whisper strategy' aimed at turning other academics against social psychologist Jonathan Haidt, a vocal critic linking social media to teen mental health problems. The notes, dated around a Substack post by Haidt, appear informal and typo-ridden, and it remains unclear whether Meta ever acted on the ideas discussed.
theverge.com
· 2026-10-28
A Show HN post introduces Screen Memories (SCM), a macOS application that lets users search photos and every frame of video using natural-language queries, OCR, and spoken-dialogue search via Whisper. All processing runs locally on-device through a local vision model, with no cloud uploads or accounts required, and the app auto-imports watched folders and re-embeds content when models change.
github.com
· 2026-10-04
According to the New York Post, women in the tech industry are circulating a private, informally shared spreadsheet listing more than 100 men—including prominent investors and businessmen—accused of sexual misconduct ranging from harassment to drugging and rape. The document is not public and is distributed selectively among women who know each other, functioning as a warning system rather than a formal complaint process.
futurism.com
· 2026-09-27
A developer created a set of Claude Code skills that convert a single chess game into a readable post-mortem, combining Stockfish engine analysis with the player's own recorded thoughts during the match. In a demonstration, Claude transcribed a French audio recording of the player's in-game reasoning via whisper.cpp, aligned it with move timestamps from the PGN, and produced a narrated video explaining specific decisions, such as bishop placement choices, verified against engine lines. All underlying files—transcript, engine analysis, annotated PGN, and storyboard—are shared in the project's example folder.
github.com
· 2026-09-26
Sources tell The Post that a private, closely guarded spreadsheet circulating among Silicon Valley women names more than 100 alleged sexual predators, including prominent investors and businessmen. One contributor, a tech worker in her 20s, said she was invited to add her own account after mentioning it privately, and that the list includes allegations of drugging, assault and rape tied to figures at major tech and AI firms. The Post notes the authenticity of individual entries has not been independently verified.
nypost.com
· 2026-09-25
European startup Desert Ant Labs has released a suite of 18 small specialized machine learning models—12 stable and six in beta—for tasks like transcription, audio cleanup, PII redaction, and language identification, all designed to run locally on devices rather than through cloud APIs. The models are accessible through a single SDK covering Swift, Kotlin, and JavaScript, and are free to use for up to 100,000 monthly active devices.
desertant.com
· 2026-09-09