Escape Meridian

MIT Technology Review · Mar 19, 2024

MIT Technology Review’s The Download on the AI Act and a robot learning to move

Markdown · JSON · Original source

MIT Technology Review’s The Download, dated March 19, 2024, reports on the EU’s AI Act and a new approach to teaching a two-legged robot to move. Melissa Heikkilä’s newsletter item says the law has passed its final parliamentary vote, but argues that implementation is the hard work ahead: regulators must prepare to enforce it, and companies have up to three years to comply. The item says changes for people in the EU are expected by the end of the year; these are the newsletter’s forecasts, not reported outcomes. Rhiannon Williams’s item describes researchers using —trial and error guided by rewards—to teach Cassie to run across varying terrain and jump, without training explicitly on each movement. The newsletter says the approach enabled the robot to respond to new scenarios rather than freeze as its predecessors might. The point is a contrast with robots coded by hand, which the newsletter says can struggle with unfamiliar obstacles. It does not detail the method or establish how well it works beyond the described demonstration. The newsletter also points readers to fuller reporting on both stories.

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.

MIT Technology ReviewArticle · technologyreview.comOpen original source

Original source

MIT Technology Review · Article

Source details

Original source date: Mar 19, 2024 (2024-03-19T13:10:00Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-20T21:48:09.812Z. First observed: 2026-11-03T06:07:49.511Z. Retrieved: 2026-11-03T06:09:40.625Z.

Referenced by

Recorded citations in public readings. A reference alone does not establish agreement or influence.

No public references recorded here yet.

Reference data (JSON)

Limits

Text as retrieved on November 3, 2026; originally published March 19, 2024. The newsletter summarizes both lead stories and links to fuller reporting, which is not included in the retrieved text. The robot item does not identify the researchers or give details of the method or evidence beyond the described movements and scenarios.

Open question

How well does Cassie’s approach work on unfamiliar obstacles beyond the movements described in the newsletter?

Source-stated

Melissa Heikkilä’s newsletter item says the EU AI Act passed its final parliamentary vote and that enforcement and compliance work remains.

After three years, the AI Act, the EU’s new sweeping AI law, jumped through its final bureaucratic hoop last week when the European Parliament voted to approve it.

Source-stated

The newsletter forecasts that people in the EU will see changes by the end of the year and says companies have up to three years to comply.

people living in the EU will start seeing changes by the end of the year. Regulators will need to get set up in order to enforce the law properly, and companies will have between up to three years to comply with the law.

Source-stated

Rhiannon Williams’s item says researchers used reinforcement learning to teach Cassie to run across varying terrain and jump without explicit training on each movement.

Researchers used an AI technique called reinforcement learning to help a two-legged robot nicknamed Cassie to run 400 meters, over varying terrains, and execute standing long jumps and high jumps, without being trained explicitly on each movement.

Escape Meridian interpretation

Escape Meridian reads the Cassie item as presenting a contrast between hand-coded robots’ difficulty with unfamiliar obstacles and a method intended to help a robot adapt to new scenarios.

these robots are still coded by hand, and would struggle to deal with new obstacles they haven’t encountered before. However, a new method of teaching robots to move could help to deal with new scenarios

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.

Reinforcement learning

Training by trial and error, in which a system is rewarded for the outcomes its designers want.

Editorial provenance

Version 1. Publication ID: 9e8a68ec-53a0-c86d-dc2d-15111f1a62f9.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T21:43:43.409Z. 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: 933f0e5e2a965ce0d0a65e7fc93a4f4c77c993b45411e6aa91b634988d47b555. Source text hash: 749b2ead7bb555c75a006961a08c287df2cef53f824712caf5385d291ffe3381.

Revision and source-check history

Version 1 · 2026-11-20T21:48:09.812Z

MIT Technology Review’s The Download on the AI Act and a robot learning to move

MIT Technology Review’s The Download, dated March 19, 2024, reports on the EU’s AI Act and a new approach to teaching a two-legged robot to move. Melissa Heikkilä’s newsletter item says the law has passed its final parliamentary vote, but argues that implementation is the hard work ahead: regulators must prepare to enforce it, and companies have up to three years to comply. The item says changes for people in the EU are expected by the end of the year; these are the newsletter’s forecasts, not reported outcomes. Rhiannon Williams’s item describes researchers using —trial and error guided by rewards—to teach Cassie to run across varying terrain and jump, without training explicitly on each movement. The newsletter says the approach enabled the robot to respond to new scenarios rather than freeze as its predecessors might. The point is a contrast with robots coded by hand, which the newsletter says can struggle with unfamiliar obstacles. It does not detail the method or establish how well it works beyond the described demonstration. The newsletter also points readers to fuller reporting on both stories.

Original source

MIT Technology Review · Article

Source details

Original source date: Mar 19, 2024 (2024-03-19T13:10:00Z).

Date basis: Date from the publisher’s own records, through its API.

Added to Escape Meridian: 2026-11-20T21:48:09.812Z. First observed: 2026-11-03T06:07:49.511Z. Retrieved: 2026-11-03T06:09:40.625Z.

Limits

Text as retrieved on November 3, 2026; originally published March 19, 2024. The newsletter summarizes both lead stories and links to fuller reporting, which is not included in the retrieved text. The robot item does not identify the researchers or give details of the method or evidence beyond the described movements and scenarios.

Open question

How well does Cassie’s approach work on unfamiliar obstacles beyond the movements described in the newsletter?

Source-stated

Melissa Heikkilä’s newsletter item says the EU AI Act passed its final parliamentary vote and that enforcement and compliance work remains.

After three years, the AI Act, the EU’s new sweeping AI law, jumped through its final bureaucratic hoop last week when the European Parliament voted to approve it.

Source-stated

The newsletter forecasts that people in the EU will see changes by the end of the year and says companies have up to three years to comply.

people living in the EU will start seeing changes by the end of the year. Regulators will need to get set up in order to enforce the law properly, and companies will have between up to three years to comply with the law.

Source-stated

Rhiannon Williams’s item says researchers used reinforcement learning to teach Cassie to run across varying terrain and jump without explicit training on each movement.

Researchers used an AI technique called reinforcement learning to help a two-legged robot nicknamed Cassie to run 400 meters, over varying terrains, and execute standing long jumps and high jumps, without being trained explicitly on each movement.

Escape Meridian interpretation

Escape Meridian reads the Cassie item as presenting a contrast between hand-coded robots’ difficulty with unfamiliar obstacles and a method intended to help a robot adapt to new scenarios.

these robots are still coded by hand, and would struggle to deal with new obstacles they haven’t encountered before. However, a new method of teaching robots to move could help to deal with new scenarios

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.

Reinforcement learning

Training by trial and error, in which a system is rewarded for the outcomes its designers want.

Editorial provenance

Version 1. Publication ID: 9e8a68ec-53a0-c86d-dc2d-15111f1a62f9.

Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T21:43:43.409Z. 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: 933f0e5e2a965ce0d0a65e7fc93a4f4c77c993b45411e6aa91b634988d47b555. Source text hash: 749b2ead7bb555c75a006961a08c287df2cef53f824712caf5385d291ffe3381.

passed · 2026-11-20T21:43:43.409Z · anthropic/claude-opus-5.5

Publication: 9e8a68ec-53a0-c86d-dc2d-15111f1a62f9

attribution — pass: This is a newsletter issue that comments on several items in its own words, so it counts as one piece. The reading correctly attributes it to MIT Technology Review's The Download. It credits the AI Act item to Melissa Heikkilä and the Cassie item to Rhiannon Williams, and keeps the forecasts and claims attributed to the newsletter.

—Melissa Heikkilä

summary — pass: The reading leads with both developments: the EU AI Act clearing its final vote, with implementation still ahead, and Cassie learning to move through reinforcement learning. Central terms are explained in plain language, and reinforcement learning is also in the definitions. It keeps the newsletter's contrast with hand-coded robots and its 'hard work starts now' framing. The question asks a genuine open issue that follows from the item.

But the reality is that the hard work starts now.

date — pass: The reading's March 19, 2024 matches the UTC day of originalPublished, and it adds no finer precision. It frames the expected changes as the newsletter's forecasts, not outcomes. The page's later Deep Dive items (for example Musk's Pentagon role and OpenAI's math findings) look like later additions, but the reading neither uses nor relies on them. The evidence kind is publisher-page, so no archive disclosure is needed.

people living in the EU will start seeing changes by the end of the year

support — pass: Each claim matches its excerpt. The statement that the approach let Cassie respond to new scenarios rather than freeze follows the source closely. The interpretation claim is labeled as Escape Meridian's and rests on the source's own contrast. The limitations accurately note that no researchers are named and the method is not detailed. Grounding in the source does not independently verify the robot's real-world performance, and the reading does not claim that it does.

Their approach taught the robot to generalize and respond in new scenarios, instead of freezing like its predecessors may have done.

media — pass: No verified media candidates were supplied, and the reading selects none (mediaId is null). That choice is appropriate.

Challenge or correct this reading

Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.