---
title: "Technocrats vs. Biological Threat: AI Leaders Demand Strict Control Over DNA Synthesis"
description: "Leaders of OpenAI, Google DeepMind, and Anthropic have called on the US Congress for strict control over DNA synthesis. AI has learned to bypass biolab security and assist in the creation of biological weapons. 🧬🚫"
date: 2026-06-05T05:14:00.000Z
lang: en
url: https://xab.info/en/posts/technocrats-vs-biological-threat-ai-leaders-demand-strict-control-over-dna-synthesis
tags: []
publisher: "XAB.info"
---

# Technocrats vs. Biological Threat: AI Leaders Demand Strict Control Over DNA Synthesis

![3D model of a DNA double helix in blue and white tones, symbolizing biological threat and the need for control over genetic material synthesis](https://xab.info/media/2026/06/05/lideryi-ii-trebuyut-zhestkogo-kontrolya-za-sintezyom-dnk/lideryi-ii-trebuyut-zhestkogo-kontrolya-za-sintezyom-dnk-1.webp)

Leaders of the world's largest artificial intelligence companies — OpenAI, Anthropic, and Google DeepMind — have joined forces to warn of a global threat. They signed an open letter addressed to the US Congress, calling for strict legislative measures to control DNA synthesis. The main goal of the initiative is to prevent the possibility of creating biological weapons using powerful neural networks.

### AI as a Catalyst for Biological Danger

Modern technology has reached a point where the creation of genetic sequences has become an accessible and automated process. Dozens of commercial companies around the world offer services to "print" custom DNA for scientific and medical purposes. However, as experts note, not all suppliers conduct thorough checks of their clients and orders.

The situation is exacerbated by the emergence of powerful language models and AI tools for protein design. Artificial intelligence can not only accelerate the process but also help malicious actors bypass existing security measures in three ways:

    - **Finding Loopholes:** AI can instantly suggest to the user which DNA suppliers to contact to avoid checks.

    - **Order Masking:** Neural networks can rephrase or alter the structure of a genetic order so that the laboratory's protective software does not recognize the toxicity or viral nature of the code.

    - **Generation of New Threats:** Models are capable of creating entirely new dangerous sequences that have the structure of known toxins but are completely ignored by standard filtering software.

### A History of Warning

Risks associated with the accessibility of DNA synthesis are not theoretical. As early as 2017, researchers were able to order necessary fragments by mail and recreate the extinct horsepox virus — a close relative of the deadly smallpox virus. Since then, synthesis technology has become significantly cheaper, making it accessible to a wide range of people.

Today, the automatic security systems of biotech laboratories often fail to recognize modified sequences of dangerous pathogens created using algorithms.

### Who is Behind the Initiative

The initiators of the open letter were the Institute for Progress and the Foundation for American Innovation. Among the signatories are key industry figures: Demis Hassabis (Google DeepMind), Sam Altman (OpenAI), Dario Amodei (Anthropic), and Mustafa Suleyman (Microsoft AI), as well as a number of leading scientists.

AI industry leaders are urging Congress to require all synthetic DNA and RNA suppliers to conduct mandatory screening of clients and orders for dangerous pathogens. Currently, in the US, there are only recommendations for institutions with federal funding, while the general market relies on voluntary audits within the International Gene Synthesis Consortium.

### Legislative Response and Internal Restrictions

In response to growing concerns, a bipartisan bill is already being considered in the Senate that should introduce mandatory checks at the legislative level. However, biosecurity experts warn that control only at the DNA printing stage is not a panacea, as the filters of automatic protection systems are imperfect.

Alongside legislative measures, AI laboratories must implement their own internal restrictions. Models must be designed to block any user requests related to the development of toxins or pathogens capable of triggering a global pandemic.