Google has staked a significant part of its future on Gemini, its flagship family of AI models. Some of the people building it, however, have reportedly been making fun of it.

A June 4, 2026 report found that certain Google engineers have turned to memes to lampoon the company's internal AI tools. Their complaints include data fabrication, burdensome code reviews, and multi-month delays on key models — an awkward look for a company trying to convince the world that its AI is ready for serious work.

What the engineers are complaining about

The report points to three recurring frustrations. The first is data fabrication, meaning the tools can produce information that is not real — a failure mode widely known in the industry as hallucination, and one of the most persistent challenges for generative AI systems. The second is code reviews that engineers describe as a burden rather than a help; AI-assisted code review has become a routine part of developer workflows across the tech industry, so its reliability matters to the people using it every day.

The third, and probably the most consequential, is the timing of Gemini 3.5 Pro. The model's launch was delayed over several months, and its coding performance reportedly fell below Google's own internal benchmarks. Coding ability has become one of the most closely watched measures of a frontier model's usefulness, since it directly shapes whether enterprises and developers judge a model ready for production work.

According to the report, the discontent is being fueled by anxiety over practical usability and pressure to implement the tools, rather than by any single controversial model announcement. The absence of a single flashpoint has likely muted any collective outcry, the report suggests.

Strong commercial numbers tell a different story

The internal friction stands in contrast to Gemini's commercial performance. Gemini models have reportedly reached approximately 900 million monthly users, and Google saw a 16% increase in ad revenue over the same assessment period. A user base of that scale places Gemini among the most widely used AI products available. Gemini is also central to the company's strategic framework under CEO Sundar Pichai.

Public coverage of the internal mockery has been limited, with few major outlets reporting on it. That may partly reflect how diffuse the complaints are, spread across tooling and workflow issues rather than attached to one headline event.

A familiar pattern for Gemini

This is not Gemini's first rough patch. Earlier versions of the family faced both public and internal criticism following their launches in 2023 and 2024. Those earlier complaints centered on bias and on performance that lagged expectations, while the current round of skepticism focuses more on day-to-day reliability and delivery timelines.

The report also notes that internal feedback loops at Google serve a dual role: they function as criticism, but also act as a mechanism for improving the AI tools.

What this means for Google

A multi-month delay on a flagship model like Gemini 3.5 Pro is not a small matter in a field where release cadence shapes perception, and where developers and customers track shipping schedules closely. Falling short of internal coding benchmarks raises the question of what Google will need to change before the model ships.

The report describes implementation pressure as another factor. When companies push staff to adopt AI tools quickly, adoption can outpace the tools' readiness — a tension that has become familiar across the tech industry as generative AI becomes embedded in everyday workplace tooling.

The reported approximately 900 million monthly users and 16% ad revenue growth give Google room to work through these issues. One thing to watch is when Gemini 3.5 Pro actually arrives, and whether its coding performance closes the gap with internal benchmarks. Another is whether the lack of a single contentious announcement continues to keep the discontent diffuse, or whether a high-profile misstep turns scattered memes into a louder and more organized conversation.