Experts push for DSM-6 to formally recognize algorithmic environments as a factor in eating disorders
Original reporting: STAT News
A group of experts writing in STAT News argues that the next edition of the DSM, expected around 2030, should explicitly account for the role digital algorithms play in shaping the environments of patients with eating disorders. The call reflects growing clinical concern that recommendation engines and content feeds are not incidental to these conditions but may be actively reinforcing them.
Why it matters
The argument that diagnostic manuals should reflect the environments patients actually inhabit is not radical. The DSM has evolved before in response to cultural and contextual shifts. What is different here is that the environment in question is engineered, personalized, and largely invisible to the clinician in the room. A patient can describe what they eat and what they feel, but they rarely describe what their feed served them at 11pm the night before.
The clinical instinct behind this proposal is sound. The implementation challenge is significant. Recommending that DSM-6 address algorithmic environments is one thing. Building the assessment tools, the clinical interview frameworks, and the evidence base that would make that criterion actionable is another task entirely, and the field is not close to having it. The 2030 timeline gives some runway, but only if the research community treats this as an empirical problem to solve rather than an advocacy position to hold.
The ReasonFirst take
Getting algorithmic exposure into the DSM is worth pursuing, but the harder clinical problem is that we do not yet have reliable, validated ways to measure what a given patient's feed is actually doing to them, and a diagnostic label without that measurement infrastructure risks being more symbolic than useful.
Who should care
What to watch
Whether the DSM-6 working groups convened by APA formally take up digital environment criteria, and whether any tech platform data becomes accessible enough to support that kind of diagnostic evidence base.
A question worth sitting with
If the DSM acknowledges algorithmic harm as a clinical variable, what obligation does that create for platforms, and who enforces it?
More signals
FDA's TEMPO Pilot Lets Some AI Medical Devices Reach Patients Before Full Authorization
STAT News · September 3, 2026
Trump Administration Reaches Drug Pricing Deals With Manufacturers: What We Know and What We Don't
STAT News · September 1, 2026
House Stopgap Bill Temporarily Blocks White House Bid to Put Political Appointees in Charge of Research Grants
STAT News · September 1, 2026