Autism diagnoses have risen dramatically over the past few decades, especially at the milder end of the spectrum—often called “Level 1” or “high-functioning” autism.

Many families and adults diagnosed later in life experience this change as a blessing: People who might once have been overlooked or marginalized are finally being recognized.

But there is another possibility. Some clinicians and researchers fear that criteria for milder autism have stretched so far that they now include people whose social difficulties stem more from anxiety, emotional sensitivity, awkwardness, or high intelligence with niche interests rather than from autism itself.

That tension sits at the heart of recent comments by Dame Uta Frith. Frith is an emeritus professor of cognitive development at University College London and a pioneering autism researcher whose work dates back to the 1960s. Her contributions helped shape our modern understanding of autism as a neurodevelopmental condition.

In an interview with TES Magazine, Frith argued that the autism spectrum has become so broad that it is “close to collapse” and has “lost all meaning” as a medical category (Amass, 2026).

The Concern: A Diagnosis That Is Expanding Rapidly

Frith’s central argument is that the autism spectrum has become too broad. She notes that the label is increasingly applied to people with a wide range of social difficulties or sensory sensitivities. In her view, under the banner of inclusivity and self-identification, autism risks becoming a less precise clinical construct and more a vague identity category—often amplified by social media.

The reaction to Frith from parts of the autism community has been swift and fierce. Critics have accused Frith of being “out of touch,” of wanting to “send autism back to the dark ages,” and discounting her expertise because she ignores their “lived experience.”

Yet Frith is far from the first researcher to raise concerns about overdiagnosis.

Evidence That Some Diagnoses May Not Hold Over Time

Direct evidence for autism overdiagnosis comes from studies looking at diagnostic stability. A large U.S. survey analyzed by the Centers for Disease Control and Prevention found that about 13 percent of children who had once received an autism diagnosis were later reported by their parents to have “lost” it. In most of those cases, parents said the change occurred because new information—or a different clinician—concluded that the child likely did not have autism after all, rather than because the symptoms had simply improved (Blumberg et al., 2016).

Other research suggests that, at least in some settings, diagnostic practices may also play a role. For example, clinical observations reported in one study described situations in which children with developmental delays or unusual behaviors were labeled with autism even though they did not fully meet the criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). The authors suggest that variability in clinician training and experience can contribute to this kind of diagnostic inaccuracy (Fegundes & Goulie, 2025).

In another study, researchers revisited children who had previously received a community diagnosis of autism and conducted comprehensive reassessments using standardized testing and expert clinical review. After the re-evaluations, about 23 percent of the children were classified as “non-spectrum,” suggesting that a meaningful minority of earlier diagnoses did not meet stricter research standards upon closer examination (Hodges et al., 2018).

None of this means that autism is widely misdiagnosed. But it does suggest that, as the diagnostic net has widened, some degree of over-classification has occurred.

Multiple Researchers Raising Similar Concerns

Several other researchers and commentators have noted similar trends.

For example, Fombonne (2025) reviewed prevalence data and diagnostic practices in an editorial in the Journal of Child Psychology and Psychiatry. His conclusion was cautious but clear: While many factors contribute to rising autism rates, overdiagnosis—particularly in milder cases—may account for at least part of the increase.

Likewise, Allen Frances, who led the task force responsible for the DSM-IV, has expressed regret about how broadly autism diagnoses have expanded over time. In a 2023 statement, Frances described the widening of diagnostic boundaries as his “biggest DSM-IV regret,” arguing that it contributed to what he believes has been substantial overdiagnosis.

A View From the Clinic

These debates are not just theoretical. Increasingly, in our clinic, many children and teens who present for treatment have previously been diagnosed with autism. However, even a cursory review of the DSM-5 diagnostic criteria, especially the clear failure to meet all three (or any) components of Criterion A: Social communication/interaction, leaves us scratching our heads. And it’s often not even a close call; I don’t think any reasonable person would agree the kid meets DSM criteria.

Occasionally, the discrepancy is striking enough to bring to mind Gertrude Stein’s famous remark: “There is no there, there.”

As Any DBT Therapist Will Tell You Two Things Can Be True at the Same Time

The temptation in debates like this is to frame the issue as a stark choice: either the expansion of autism diagnoses is entirely beneficial, or it represents a serious diagnostic problem.

Reality is more complicated. Despite the temptation to dichotomize this issue and pick a side, both can be true simultaneously: broadening criteria can bring substantial benefit to some while still potentially harming others.

Asking the Question Should Not Be Taboo

The broadening of diagnostic criteria has also brought meaningful benefits to many. Expanding the definition of autism has made it more likely that individuals who might previously have been overlooked are now identified and able to access appropriate supports. This represents important progress toward more equitable inclusion for those whose needs were historically marginalized and invalidated.

However, despite the criticism—and sometimes outright vitriol—that Frith has experienced, raising the question of overdiagnosis should not be grounds for vilification or an attempt by one side to shut down the conversation.

Scientific progress depends on asking difficult questions. And it’s not just an academic exercise; our patients are relying on us to get this right.

Perhaps future evidence will show that Frith and others who share her concerns are mistaken. But no researcher deserves to be crucified simply for asking questions.