Dia (Product): news timeline & CVEs

Bring Your Own Agent: Your Employees' Assistants Are Already Logged In

Employees are linking personal AI agents to corporate apps through OAuth, passwords, and API tokens, creating unmanaged identity risk.

Personal AI agents including Meta's Muse, OpenAI's Dots, xAI's GrokBot, Instinct, and open-source OpenClaw are being connected to workplace systems by employees, often without IT tickets. Access commonly comes through OAuth consent to Google Workspace, Microsoft 365, or Slack, saved browser passwords, API keys and personal access tokens, or local agents on employee laptops. Vendors are adding controls only for managed corporate deployments, so personal agents do not appear in admin consoles. Instinct, a text-based agent, raised $1 billion at a $10 billion valuation and has drawn privacy concerns.

The Hacker NewsOct 12, 11:55 UTC · 4d agoAI safety & security 2 sources6· 4 reads

PPO-HRAP: Proximal Policy Optimization with a Hybrid Regime-Aware Policy for Risk-Controlled Trading

PPO-HRAP blends a regime prior with PPO, cutting SPY max drawdown to 18.47% versus 34.10% buy-and-hold.

PPO-HRAP combines Proximal Policy Optimization with an interpretable regime prior for risk-controlled trading. Rewards mix portfolio log return, a VIX-conditioned drawdown penalty, target-exposure deviation, and turnover cost, while actions blend the actor with a regime-derived exposure. On the 2020-2022 SPY window it reports 27.62% total return, 8.48% annualized return, Sharpe 0.6447, and maximum drawdown of 18.47% versus 34.10% for buy-and-hold. Five SPY seeds average total return 0.2725 and Sharpe 0.6219; single-run QQQ and DIA tests also rank first, though turnover remains high.

arXiv cs.AI / cs.LG / cs.CLOct 1, 08:49 UTC · 15d agoAI research1

Dia

Timeline2 articles

Appears with

Bring Your Own Agent: Your Employees' Assistants Are Already Logged In

PPO-HRAP: Proximal Policy Optimization with a Hybrid Regime-Aware Policy for Risk-Controlled Trading