35% of a Berkeley CS class failed — and cheating is only part of it

A 35% failure rate, and a professor who blames the admissions office

The Daily Californian reported a number that does not happen in introductory CS: in spring 2026, 35.3% of CS 10 students received F's, against under 10% in prior years. CS 61A — the flagship intro course — failed 10.6%. Upper-division EECS 127 failed 16.8%, where 5% is normal. Berkeley's own guidelines say a typical lower-division GPA should sit between 2.8 and 3.3; both intro classes averaged a C-plus (2.3).

Professor Dan Garcia, who teaches CS 10 and CS 61A, ties it to surging AI use and students arriving without the linear algebra they need. Professor Gireeja Ranade notes courses now run with an "open-internet, open-AI policy" on assignments and exams. The blunt subtext: students who let the model do the homework hit a wall when the exam takes the model away.

The detail buried in the article that flips the headline

The story's own paragraphs complicate the AI narrative — and the thread found them fast.

"Some of the numbers that you saw from the number of students who receive failing grades were because we caught them (cheating)… According to Garcia, nearly 30 students in CS 10 were caught cheating on take-home exams." — a commenter quoting the article

So a chunk of the F-spike is enforcement, not aptitude. The deeper cause may sit in admissions, not AI:

"The likely 'real' reason is hidden in one paragraph… Both Garcia and Ranade have joined more than 1,300 UC faculty in signing a petition calling for the reinstatement of ACT and SAT… It's been a failure everywhere, with many, if not most, universities already reversing it." — a reader citing Yale's data that test scores are 'the single greatest predictor' of grades

And the math erosion long predates LLMs — by an account from inside CS teaching:

"My first math exam as a CS undergraduate, 123 out of 129 students failed… Math was core to the CS curriculum in those days. It would fade away over the next few decades to almost nothing. The main reason being the CS department wanted to popularize its uptake, and remove barriers." — a former CS instructor

A contrarian read: the failures may be the system finally telling the truth.

"Professors suddenly realized everyone was cheating and started paying attention, but the cheating isn't new… AI woke some of them up to reality." — a commenter who argued the cheating predates AI

And one Cal alum welcomed it:

"As a Cal alum, I am actually really glad to see they are holding the line on grade inflation… it would really suck to see my labor devalued if Cal went the direction of e.g. Yale." — a Cal alum

Even the worry about adults isn't theoretical for some:

"Now I work mostly with PhDs who were at the top of every academic environment… And yet I can see their thinking skills rapidly declining as well; many of them can no longer brainstorm, code, think deeply, or write without an LLM present doing 90% of the work." — an engineer working with PhD colleagues

Three causes wearing one headline

The failure spike isn't one story but three braided together: a cheating crackdown that converted undetected cheating into recorded F's, six years of test-optional admissions letting in students without the prerequisites, and an AI workflow that lets you ship homework you never understood. Strip out any single one and the number still moves. The professors holding the line on grades did the unusual thing — they let the F's stand instead of curving them away — which is precisely why the rest of us got to see the number at all.