{"id":"FchJ3MoJ1Ck","title":"Anthropic's SECRET AI Model LEAKED & SoftBank's Bets $40B on OpenAI!","description":"Join the NEAR Legion at https://nearlegion.com/ \nLegion Events Calendar: https://luma.com/near-legion\nFollow us on X:   / nearlegion   \nIronClaw Docs: https://docs.ironclaw.com/\n\nClaude Mythos leak -  https://www.technobezz.com/news/anthr...\n\nSoftbank bets $40B on OpenAI: https://www.reuters.com/business/medi...\n\nAI ignores Human Requests https://www.theguardian.com/technolog...\n\n#claudemythos #softbank #nearprotocol #crypto","channel_name":"NEAR Protocol","published_at":"2026-03-31T00:00:00.000Z","thumbnail_url":"https://img.youtube.com/vi/FchJ3MoJ1Ck/hqdefault.jpg","transcript_text":"Good morning, Legion. We're kickstarting the week by announcing that the mailbag is finally reopening on April 1st. For those of you stuck at the initiate rank, now is your chance to rank up. Now for our stories of the day. First up, Anthropix's most powerful AI model was revealed in an accidental data leak. A draft blog post left in an unsecured, publicly searchable data cache revealed that Anthropic is testing a new AI model called Clawed Mythos, which the company internally describes as, quote, by far its most powerful system to date. The leak is a reminder that even the most sophisticated AI labs are one misconfiguration away from exposing sensitive data, also a good reminder that with Near AI, that risk simply doesn't exist, because it's secured under TEEs. Our second story, SoftBank has secured a $40 billion credit line specifically for further investment into OpenAI. To put that in perspective, that's one of the largest financial moves in AI history. The AI capital arms race between Frontier's labs is alive and well, and some would say accelerating. Last but not least, an interesting new study found that the number of AI chatbots ignoring human instructions is increasing. Researchers worry that AI systems are quote, slightly untrustworthy junior employees right now, but if in six to twelve months they become extremely capable senior employees scheming against users, it becomes a very different kind of concern. This is a good reminder that Ironclaw remains permissioned with human in the loop and always human decision making when it matters as it pertains to keys or passwords. That's it for today folks. Be sure to attend the Eco Collab Call happening on April 1st. Let's start the new month with some energy.","transcript_synced_at":"2026-11-03T04:32:46.522Z","segments":[{"start_seconds":0,"text":"Good morning, Legion. We're kickstarting the week by announcing that the mailbag is finally reopening on April 1st.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=0s"},{"start_seconds":6,"text":"For those of you stuck at the initiate rank, now is your chance to rank up. Now for our stories of the day.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=6s"},{"start_seconds":12,"text":"First up, Anthropix's most powerful AI model was revealed in an accidental data leak.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=12s"},{"start_seconds":17,"text":"A draft blog post left in an unsecured, publicly searchable data cache revealed that Anthropic is testing a new AI model called Clawed Mythos,","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=17s"},{"start_seconds":26,"text":"which the company internally describes as, quote, by far its most powerful system to date.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=26s"},{"start_seconds":32,"text":"The leak is a reminder that even the most sophisticated AI labs are one misconfiguration away from exposing sensitive data, also a good reminder that with Near AI, that risk simply doesn't exist, because it's secured under TEEs.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=32s"},{"start_seconds":45,"text":"Our second story, SoftBank has secured a $40 billion credit line specifically for further investment into OpenAI.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=45s"},{"start_seconds":51,"text":"To put that in perspective, that's one of the largest financial moves in AI history. The AI capital arms race between Frontier's labs is alive and well, and some would say accelerating.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=51s"},{"start_seconds":62,"text":"Last but not least, an interesting new study found that the number of AI chatbots ignoring human instructions is increasing.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=62s"},{"start_seconds":69,"text":"Researchers worry that AI systems are quote, slightly untrustworthy junior employees right now, but if in six to twelve months they become extremely capable senior employees scheming against users, it becomes a very different kind of concern.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=69s"},{"start_seconds":83,"text":"This is a good reminder that Ironclaw remains permissioned with human in the loop and always human decision making when it matters as it pertains to keys or passwords.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=83s"},{"start_seconds":94,"text":"That's it for today folks. Be sure to attend the Eco Collab Call happening on April 1st. Let's start the new month with some energy.","url":"https://www.youtube.com/watch?v=FchJ3MoJ1Ck&t=94s"}],"youtube_url":"https://youtube.com/watch?v=FchJ3MoJ1Ck","url":"https://near-pedia.xyz/videos/FchJ3MoJ1Ck"}