ZeroDrift launched Command, a compliance platform that automatically detects and corrects regulatory violations in emails and AI-generated messages before they are sent.
The platform uses a specialized small language model to inspect communications in real time against regulatory requirements and company policies, addressing the compliance challenge created by AI agents that generate high volumes of messages autonomously.
Financial regulators hold companies responsible for AI communications as they would employee messages, making real-time inspection increasingly critical for firms deploying autonomous AI agents.
What happened
ZeroDrift Inc. launched Command, a control plane that detects and corrects regulatory violations in emails and AI-generated communications before delivery. The platform applies corporate policies and industry regulations to flag issues like promissory claims, unauthorized disclosures, and references to material nonpublic information, and can automatically rewrite AI-generated content to comply or block messages entirely.
Why it matters
As companies deploy AI agents that generate thousands of messages in seconds, manual compliance review is becoming impractical. Financial regulators hold companies responsible for AI-generated communications the same way they do for employee-written messages. Command inspects communications in real time rather than after delivery, reducing compliance risk at scale.
What to watch
ZeroDrift's Anchor small language model detects violations in milliseconds (under one-second latency), compared to over 10 seconds for some general-purpose models like Anthropic's Claude. The company plans to extend coverage to WhatsApp, SMS, Slack, and Microsoft Teams. Pricing is based on input and output tokens processed, but details have not been disclosed.
Ask the AI about this article →
ZeroDrift's launch of Command addresses a fundamental tension in modern compliance: the speed of AI-generated communications now vastly outpaces the human review processes designed to catch violations. Financial services regulators treat AI-generated messages with the same legal accountability as employee communications, yet traditional compliance systems that review messages after delivery or require manual intervention are impractical when a single AI agent can produce thousands of messages in seconds. The startup's approach—intercepting communications in real time and applying deterministic checks (keywords, pattern matching) alongside a specialized small language model trained on regulatory data—positions the platform as fundamentally different from post-hoc review.
The performance gains ZeroDrift claims are striking. The company reports that its Anchor 3.0 model detects at least twice as many violations as general-purpose frontier models and operates in milliseconds rather than the 10+ seconds required by approaches using Claude or similar large language models. These latency improvements matter because they determine whether real-time compliance inspection is operationally feasible at scale. However, ZeroDrift explicitly cautions that Command is a risk-reduction tool rather than a guarantee; the company is still in a research phase evaluating Anchor against 100,000 human-annotated emails and has not yet disclosed detailed detection rates or false-positive results.
The expansion roadmap—adding WhatsApp, SMS, Slack, and Microsoft Teams to the current email and AI agent output coverage—suggests the startup sees compliance violations as endemic to wherever autonomous communication happens, not limited to formal channels. Pricing based on token volume ties cost directly to the scale of communication, which could align incentives but also creates a variable cost model that will depend on how token usage evolves as customers deploy more AI agents.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
Visko raised $10 million in pre-seed funding from Llama Ventures and opened public access to its first foundat…
U.S. markets ended August higher, with the S&P 500 up 2.6% and the Nasdaq up 3.9%
Neurovia AI, an Abu Dhabi-based company, is pitching Saudi security agencies software that it says can compres…
AI company Runway has unveiled Solaris, the first model in a new category it calls "Interface World Models." I…
Google's AI search gave advice to call emergency services for users alone with an African, Indian, or Pakistan…
John Deere introduced JD, a conversational AI tool that lets farmers ask open-ended questions about their hist…
