MIT Technology Review · May 23, 2025
MIT Technology Review’s The Download introduces Cathy Tie and Anthropic’s new AI models
Markdown · JSON · Original source
MIT Technology Review’s May 23, 2025, edition of The Download introduces Cathy Tie and reports on ’s new AI models. In a brief preview of a story by Caiwei Chen and Antonio Regalado, the newsletter says Tie is He Jiankui’s new wife and that He credited her with the transformation of his X.com account. The edition also reports that Anthropic says trained on Opus 4 can tackle difficult tasks over extended periods and respond more usefully to instructions. As an example, it says Opus 4 created a guide for Pokémon Red while playing for more than 24 hours, compared with 45 minutes for Anthropic’s previous most powerful model. This is the company’s account of a model’s performance, not evidence here of how reliably it performs in other settings. The newsletter also previews Jessica Hamzelou’s argument that plans to limit access to covid vaccines are not all bad: she says UK boosters have been limited to vulnerable groups and that immunologists she spoke with agree the plans make sense. The edition gives these stories as short previews, with links to full articles; it does not provide the underlying reporting or the basis for Anthropic’s comparison.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
MIT Technology Review · Article
Source details
Original source date: May 23, 2025 (2025-05-23T12:10:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-13T23:13:25.991Z. First observed: 2026-10-30T14:04:31.643Z. Retrieved: 2026-10-30T14:05:51.150Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 30, 2026; originally published May 23, 2025. This edition gives brief previews and links to full stories; the evidence does not include the underlying reporting or the basis for ’s model comparison.
Open question
How does ’s comparison of Opus 4’s performance with its previous model translate to tasks beyond the Pokémon Red example?
Source-stated
MIT Technology Review’s newsletter says He Jiankui credited Cathy Tie with the transformation of his X.com account.
He emailed us back: “It’s thanks to Cathy Tie.”
Source-stated
The newsletter reports that Anthropic says AI agents trained on Claude Opus 4 tackle difficult tasks over extended periods and respond more usefully to instructions.
AI agents trained on Claude Opus 4, the company’s most powerful model to date, raise the bar for what such systems are capable of by tackling difficult tasks over extended periods of time and responding more usefully to user instructions, the company says.
Source-stated
The newsletter says Opus 4 created a guide for Pokémon Red while playing for more than 24 hours, while Anthropic’s previously most powerful model played for 45 minutes.
Opus 4 created a guide for the video game Pokémon Red while playing it for more than 24 hours straight. The company’s previously most powerful model was capable of playing for just 45 minutes.
Source-stated
Jessica Hamzelou argues that plans to limit access to covid vaccines are not all bad, citing UK booster policy and immunologists she spoke with.
In the UK, where I live, covid boosters have been offered only to vulnerable groups for a while now. And the immunologists I spoke to agree: The plans make sense.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: bda96678-9430-de7b-4671-234524148445.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-13T23:09:26.459Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: c2042b5e0e1844e836a110c59ac41228f5de5d4a0f5fb77234b5404fddf825ec. Source text hash: dec6dbd1a6c40f9145cb59a966f358b61b7046ed3dcba85bced52feff327ef55.
Revision and source-check history
Version 1 · 2026-11-13T23:13:25.991Z
MIT Technology Review’s The Download introduces Cathy Tie and Anthropic’s new AI models
MIT Technology Review’s May 23, 2025, edition of The Download introduces Cathy Tie and reports on ’s new AI models. In a brief preview of a story by Caiwei Chen and Antonio Regalado, the newsletter says Tie is He Jiankui’s new wife and that He credited her with the transformation of his X.com account. The edition also reports that Anthropic says trained on Opus 4 can tackle difficult tasks over extended periods and respond more usefully to instructions. As an example, it says Opus 4 created a guide for Pokémon Red while playing for more than 24 hours, compared with 45 minutes for Anthropic’s previous most powerful model. This is the company’s account of a model’s performance, not evidence here of how reliably it performs in other settings. The newsletter also previews Jessica Hamzelou’s argument that plans to limit access to covid vaccines are not all bad: she says UK boosters have been limited to vulnerable groups and that immunologists she spoke with agree the plans make sense. The edition gives these stories as short previews, with links to full articles; it does not provide the underlying reporting or the basis for Anthropic’s comparison.
Original source
MIT Technology Review · Article
Source details
Original source date: May 23, 2025 (2025-05-23T12:10:00Z).
Date basis: Date from the publisher’s own records, through its API.
Added to Escape Meridian: 2026-11-13T23:13:25.991Z. First observed: 2026-10-30T14:04:31.643Z. Retrieved: 2026-10-30T14:05:51.150Z.
Limits
Text as retrieved on October 30, 2026; originally published May 23, 2025. This edition gives brief previews and links to full stories; the evidence does not include the underlying reporting or the basis for ’s model comparison.
Open question
How does ’s comparison of Opus 4’s performance with its previous model translate to tasks beyond the Pokémon Red example?
Source-stated
MIT Technology Review’s newsletter says He Jiankui credited Cathy Tie with the transformation of his X.com account.
He emailed us back: “It’s thanks to Cathy Tie.”
Source-stated
The newsletter reports that Anthropic says AI agents trained on Claude Opus 4 tackle difficult tasks over extended periods and respond more usefully to instructions.
AI agents trained on Claude Opus 4, the company’s most powerful model to date, raise the bar for what such systems are capable of by tackling difficult tasks over extended periods of time and responding more usefully to user instructions, the company says.
Source-stated
The newsletter says Opus 4 created a guide for Pokémon Red while playing for more than 24 hours, while Anthropic’s previously most powerful model played for 45 minutes.
Opus 4 created a guide for the video game Pokémon Red while playing it for more than 24 hours straight. The company’s previously most powerful model was capable of playing for just 45 minutes.
Source-stated
Jessica Hamzelou argues that plans to limit access to covid vaccines are not all bad, citing UK booster policy and immunologists she spoke with.
In the UK, where I live, covid boosters have been offered only to vulnerable groups for a while now. And the immunologists I spoke to agree: The plans make sense.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
AI agent
An AI system that chooses and carries out actions toward a goal, often using software tools.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Claude
Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Editorial provenance
Version 1. Publication ID: bda96678-9430-de7b-4671-234524148445.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-13T23:09:26.459Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: c2042b5e0e1844e836a110c59ac41228f5de5d4a0f5fb77234b5404fddf825ec. Source text hash: dec6dbd1a6c40f9145cb59a966f358b61b7046ed3dcba85bced52feff327ef55.
uncertain · 2026-11-13T23:09:26.459Z · anthropic/claude-opus-5.5
Publication: bda96678-9430-de7b-4671-234524148445
attribution — pass: This is one newsletter issue that comments on several stories in its own words, so it counts as one piece. The reading attributes the edition to MIT Technology Review and correctly credits the Tie preview to Caiwei Chen and Antonio Regalado and the vaccine argument to Jessica Hamzelou. It keeps the Opus 4 capability claims as Anthropic's ('the company says').
—Caiwei Chen & Antonio Regalado
summary — uncertain: Missing essential context: the reading names He Jiankui without saying who he is. The source explains he is a Chinese biophysicist imprisoned for illegally creating the world's first gene-edited children. It also leaves 'transformation of his X.com account' vague, while the source says the account shifted to outrageous, antagonistic memes. A general reader cannot grasp why the Tie item matters. Smallest repair: add one clause identifying He Jiankui and one describing the account change. The Anthropic and vaccine items, the title and the question are clear and faithful; 'AI agent' is defined on the page.
he has sought to make a scientific comeback and to repair his reputation after a three-year incarceration for illegally creating the world’s first gene-edited children
date — pass: The reading states May 23, 2025, which matches the UTC calendar day of originalPublished, 2025-05-23T12:10Z. The page is a publisher page, so no archive disclosure is needed. The Deep Dive links that mention later items, such as OpenAI's math findings, are possible later additions; the reading neither uses nor relies on them. Anthropic's claims are reported as the company's claims, not as later outcomes.
support — pass: Each claim matches a source excerpt: He's emailed credit to Tie, Anthropic's claims about agents trained on Claude Opus 4, the 24-hour versus 45-minute Pokémon Red comparison, and Hamzelou's UK and immunologist reasoning. The caveat that the company's account does not show reliability in other settings is appropriate. The question follows from the evidence and does not assume an answer.
Opus 4 created a guide for the video game Pokémon Red while playing it for more than 24 hours straight.
media — pass: No media was selected and no verified media candidates were supplied, so the choice is consistent with the evidence.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAn AI system that chooses and carries out actions toward a goal, often using software tools.
Definition referenceClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAn AI system that chooses and carries out actions toward a goal, often using software tools.
Definition referenceClaude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.
Explore ClaudeAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore Anthropic