AI & Cybersecurity? Anthropic and OpenAI Bets Big!

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   Good morning, Legion! A couple of Near ecosystem updates to kick off the day. Near Intents is now integrated with OIC wallet built by DFINITY. OIC users can perform Intents-based cross-chain swaps directly inside the wallet, another chain and another dApp that adds Near Intents to its repertoire.

2. 0:19

   and also another one for near intense fiat to zcash is now live on peer a privacy preserving on-ramp powered by near intense no additional verification no middlemen no centralized exchange is required to use this product great enhancements from the near intense team and we're really excited about what the next couple of months is going to bring on that front as well let's get into our ai stories for the day Leading off, Anthropic launched Cloud Managed Agents, now in public beta on the Cloud Platform.

3. 0:47

   It's a hosted service that runs long-horizon agents for you, pairing a tuned agent harness with production infrastructure, so teams can go from prototype to launch in just a couple of days. The architecture decouples the agent's brain from its hands, metaphorically speaking, and its session log, meaning the harness, the sandbox, and the event store can each fail or be swapped independently.

4. 1:08

   Anthropix says this redesign dropped P50 time-to-first token by roughly 60%, then P95 by over 90%. Credentials stay outside the sandbox, so generated code never touches authentication tokens directly. If you want to build with it, the docs are live on the Cloud platform, but more importantly, this is just a first illustration of what the future of building agents might look like.

5. 1:32

   Staying with Anthropic for a moment, OpenAI is reportedly following their lead on restricting access to powerful cybersecurity models. Not surprised why, the internet probably breaks if they don't. According to Axios, OpenAI is working on a new model with advanced cyber capabilities that will only go to a small group of companies.

6. 1:50

   This mirrors Anthropic's recent move to limit access to its Mythos Preview model to select tech and security firms because of its hacking capabilities. Last but not least, Google rolled out interactive 3D models and simulations in the Gemini app.

7. 2:05

   Ask Gemini something like, show me a double pendulum or help me visualize the Doppler effect, and it generates a 3D model you can rotate, zoom, pause, and adjust with sliders in real time. The pace of AI is truly incredible as how quickly these models are updating and integrating new features.

8. 2:22

   The feature is available to all Gemini app users when you select the Pro model in the prompt bar. This comes after Anthropic gave Claw the ability to respond with charts and interactive visuals, and after OpenAI added math and science visualizations to ChatGPT.

9. 2:37

   Folks, AI is accelerating. You should follow IronClaw and join the Near Legion to keep track of how Near is currently building in this extremely fast-paced industry. And as always, follow the videos here. We'll do this all again real soon.