It is officially a White Circle Summer ⚪ At least it is in Sweden! We're excited to announce our technical partnership with Lovable. Lovable's AI agents write code, run commands, and take actions for millions of people. Most users are building great things. Some try to generate phishing sites, malware, or pull off multi-step jailbreaks. That's where White Circle comes in. AI interactions on Lovable are supervised by White Circle before an action is taken. Their team can define policies, test them against production traffic, and deploy them as new threats or failure modes emerge. That means they can respond to abuse, security threats, reliability issues, and unexpected model behavior without slowing down the product. We're proud to partner with Lovable to help make AI safer, more secure, and more reliable at scale. Learn more at https://lnkd.in/gXvY9rWf
Pre-action supervision is an important step toward making agentic systems operationally trustworthy. The ability to define policies, test them against production traffic, and update controls as new failure modes emerge creates a living governance layer rather than a static compliance checkpoint. At scale, the next requirements will include policy versioning, decision provenance, exception handling, and clear ownership when an action is blocked, escalated, or allowed. This is how safety becomes part of execution rather than an after-the-fact review.
The part worth noting is the shift from static content moderation to policy supervision that updates against live production traffic. Most AI safety layers get bolted on once and left stale as abuse patterns evolve, so a system built to redefine and redeploy policies as new failure modes emerge is a meaningfully different posture than filtering known bad inputs. The real test is how fast that loop runs when a genuinely novel jailbreak shows up, not how well it handles the attacks already in its training data.
White Circle Summer indeed ⚪ Proud of the team on this one and huge thanks to the Lovable folks for partnering with us to make agentic AI safer at scale.
🫶
See you back soon at Neon Noir. White Circle x Lovable 🖤
Lovely
Great achievement!
Critical to use safely ai agents 👌
🤝 Congrats, White Circle <> Lovable on the partnership. Making AI safer, secure & reliable at scale. ⚪️❤️
Safety is somehow being treated as part of the runtime, not as a final review. The hard question might be whether policies remain understandable to the people affected when they change as quickly as the threats.