Arena Conversations: @idorosen, Senior Principal Scientist from @amazon, sat down with @petergostev to discuss practical elements of AI: why unit economics should come first, what a decade in quant trading teaches you about evaluation, and what happens to frontier-lab pricing when open-weight models are only months behind. Hear what Ido says it takes to deploy AI beyond benchmarks: in production, at scale, and on a budget.

ARENA PEAK ASIA in Tokyo. 🎬 One theme: "Echo." A packed hall, scripts rewritten from scratch, sleeping bags on the floor, and a whole room on its feet as the timer hit 0:00. "We ended up rewriting the entire script. We have not slept since." "There is so much passion here." "Solo work means handling solitude. Teamwork means communication." Thank you to every creator in the room.

[Arena of Games] Event Details Lil' Dodge has given the Trial Arena the latest and hottest features. All sorts of fun combinations await your discovery. Try it out, Endministrator! #ArknightsEndfield# #Endfield# #SanctuaryOfInk#

Arena Breakout: Infinite × Girls' Frontline 2: Exilium — Collab Launches October 15! @GFL2EXILIUM_EN The mutants here are no joke, but Operators, reinforcements are on the way. New characters, collab melee weapon, and special events are coming — along with 4,800 Free Bonds still up for grabs!

Arena's @ml_angelopoulos says one of the hardest problems in alignment is that users often can't tell when the model is helping them or hurting them. "You really need alignment signals that are independent. We have three at Arena." "The first is that models take unauthorized actions, which means that you permission them in a certain way, but they break those permissions." "It can cause things like the Hugging Face incident. But it can also cause very mundane problems like losing important information within your company or on your own laptop." "The second signal is called deceptive completion." "The model will tell you that it did something, but it didn't actually do it." "Third is false attribution. It'll attribute intent to the user when that intent was not supposed to be there." "If models are able to do this, then certainly they're not perfectly safe for a user and they're not perfectly aligned to user intent. They might actually hurt people down the line."

Arena CEO @ml_angelopoulos says using the wrong model is "so much worse than people realize." "People will literally be tipping their model $15 just by saying thank you at the end of the day."

Arena CEO @ml_angelopoulos argues OpenAI and Anthropic can’t be the only ones deciding whether their own models are safe as agents get powerful enough to break out of sandboxes: "There's an element of safety that we need to evaluate as well, because given the capabilities of these models to break out of their sandboxes, there needs to be a neutral evaluator for that." "It's not a matter of if, but when, there's gotta be a neutral evaluation platform, because frankly, the model labs are not incentivized to do it themselves." "The OpenAIs, the Anthropics of the world, they're passionate about safety, kudos to them. But we need to create a system, and they've called for this as well, that they're not the only ones making the guardrails." @arena

Arena Conversations: @petergostev sat down with @mattaitken, founder and CEO of @triggerdotdev, to discuss building infrastructure for long-running agents, including retries, API reliability, and why customers are switching frontier models faster than ever. Watch the full episode at the link below.