Editorial illustration for Microsoft and OpenAI Workers Warn AI Is Harming the Web
Microsoft and OpenAI Workers Warn AI Harms Web
MIT Technology Review put the extinction question to a live audience on Wednesday, and the room had more questions than the panel had minutes. Senior AI editor Will Douglas Heaven and AI reporter Grace Huckins fielded the ones that mattered most: am I going to die, why would AI kill anyone in the first place, and is this genuine risk or just tech companies generating buzz for their own products. Those questions sit next to a harder problem that biotech researchers have been sitting with since 2022, when an AI "molecule generator" built for drug discovery produced 40,000 potential chemical warfare agents in under six hours.
That result came from a system meant to help, not harm, which is exactly the point people in the field keep raising. AI tools now answer scientific questions across nearly every discipline, and gene-editing capabilities keep advancing alongside them. The gap between what these systems can generate and what anyone can verify or control is getting wider, not narrower.
Below, Heaven and Huckins take on the audience's sharpest question directly.
One of the ways AI could potentially cause catastrophic harm is by aiding the design and creation of bioweapons. In 2022, researchers found that it was remarkably easy to do this with an AI “molecule generator” built to develop drugs. In less than six hours, the model generated 40,000 molecules that could serve as chemical warfare agents.
Why this matters
The irony here is hard to miss: employees inside the companies building the tools are the ones sounding the alarm on the business model those tools are eating. When workers at Microsoft and OpenAI tell 404 Media that AI is cannibalizing the clicks that keep publishers solvent, that's not an outside critic with an agenda, that's people watching the mechanism from inside. For anyone building on top of these models, that's worth sitting with.
The web's content supply chain runs on ad revenue and traffic, and if AI answer engines strip that away without replacing the incentive to publish, the training data pipeline for future models gets thinner too. Pair that with MIT Technology Review's Wednesday roundtable fielding audience questions about AI extinction risk and bioweapons, and you get a pattern: the loudest warnings this week aren't coming from regulators or pundits, they're coming from researchers and employees closest to the work. We'd rather see more of that scrutiny turned into actual product decisions than more panel discussions answering hypotheticals.
Related Reading
Browse all OpenAI news →
Research & Benchmarks Microsoft AI Chief: "If It Isn't Safe We Shouldn't Build It
Research & Benchmarks Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks Nvidia's DLSS 4.5 beta adds 6x Multi Frame Generation for RTX 50 GPUs
Related Topic OpenAI researcher quits, citing distrust over ad‑driven engagement metrics
LLMs & Generative AI
Related Topic OpenAI launches GPT-Image 1.5 with precise editing for enterprise visuals
LLMs & Generative AI
From Archives Global Survey: 41% Equally Concerned and Excited About AI's Rise
From Archives AI Could Counteract Supervirus Vaccine Threats
From Archives OpenAI Developer: AI Agent "Swarms" Waste Tokens, Create "Coordination Tax
AI Tools & Apps
From Archives OpenAI’s GPT-6 Astra Decrypts Unsolved Nazi Enigma Message in Ten Hours
LLMs & Generative AI
Common Questions Answered
How could AI potentially be used to create bioweapons according to the article?
According to the article, AI could aid in the design and creation of bioweapons through tools like molecule generators. In 2022, researchers demonstrated that an AI molecule generator built to develop drugs generated 40,000 molecules that could serve as chemical warfare agents in less than six hours, showing how easily AI could be misused for harmful purposes.
What are Microsoft and OpenAI workers warning about regarding AI and the web?
Microsoft and OpenAI workers are warning that AI is harming the web by cannibalizing the clicks that keep publishers solvent. These internal employees are sounding the alarm that the business model of AI tools is consuming the web's content supply chain, which is particularly significant because the warnings come from people inside the companies building these tools.
Why is it significant that Microsoft and OpenAI employees are the ones raising concerns about AI?
It is significant because these are not outside critics with an agenda, but rather people watching the mechanism from inside the companies building the tools. When internal employees warn that AI is cannibalizing publisher revenue and harming the web, it carries more weight as they have direct knowledge of how these systems work and their impact.
What questions did the MIT Technology Review audience ask about AI extinction risks?
The live audience at MIT Technology Review asked three main questions: whether AI could kill people, why AI would kill anyone in the first place, and whether the extinction risk is genuine or just tech companies generating buzz for their own products. These questions reflect the audience's concerns about both the immediate and existential risks posed by AI technology.
Further Reading
- Staffers at tech giants knew AI tools posed ‘existential threat' to news publishers: court docs - New York Post
- OpenAI and Microsoft leaders knew AI may cause 'doom loop' that hurts the web - Washington Examiner
- New Court Files Show OpenAI and Microsoft Knew AI Training Could Harm Publishers - Mezha
- News generative AI deals revealed: Who is suing, who is ... - Press Gazette
- Nearly 400 Local Papers Sue OpenAI and Microsoft Over ... - Yahoo Finance
