I used AI in this post. >30% is AI-generated text (everything that appears in quote blocks).

Google's Debug Project has been releasing mosquitos infected with bacteria that stop them from breeding in the wild, as a way of reducing mosquito populations.

This is a promising strategy for fighting mosquito-borne disease; Google's particular approach doesn't target malaria, but it has been shown to reduce rates of dengue fever (and similar approaches with gene drives can address malaria; the linked organization has been supported heavily by Coefficient Giving). 

The Debug Project wants to try their approach with a new species of mosquito, and they've requested permission from the EPA. If they get it, they plan to release up to 64 million mosquitos across Florida and California.

Public comments are open through June 5. Predictably, they've been mostly negative so far, because the optics of "releasing millions of infected mosquitos" are poor. (The number might be ~100%, because I read 15 comments and all 15 were negative.)

However, past releases in places like Colombia and Fresno, California show that this project is likely to be highly effective. And I'd guess that the more projects like this happen, the more it will become broadly accepted as a means of mosquito control — so this could have knock-on effects for malaria-related projects.

You can leave a comment here, or read Google's application here. I submitted a comment today (still out for review) and found the process simple.

Leaving a comment doesn't seem hugely impactful (the EPA isn't obligated to respond to them, and public opinion probably won't be a large factor here). But it is a chance to:

• Publicly support something that is beneficial both directly (by reducing disease) and indirectly (by setting a good precedent).
• Show that people in favor of large-scale mosquito control projects actually exist.
• Get some practice making public comments, and reduce the activation energy for making one if a more important opportunity comes along.

Note: I'm not an expert on these issues — just a concerned citizen who trusts the science and doesn't like the current balance of comments. If this project is actually bad, I could be missing something.

Here are notes from Claude on reasons to support the Debug Project, any of which you could cite in your comment.

Demonstrated efficacy, including by the same company in the same state

• Verily ran this exact playbook in Fresno County on a related species (Aedes aegypti). In 2018 they released 14.4 million males across three neighborhoods and the number of female mosquitoes was 95.5% lower (95% CI, 93.6–96.9) in release areas compared to non-release areas, with the most geographically isolated neighborhood reaching a 99% reduction. Across years the suppression ran 68% in 2017, 95% in 2018, and 84% in 2019. (PubMed, Hardy Diagnostics)
• The technique is validated well beyond one site. Field trials across the United States, Singapore, French Polynesia, and Australia have shown demonstrated population reductions of up to 99% (78%–99%). (NIH)
• It translates to disease, not just bug counts. In Singapore, a deployment covering ~600,000 people showed intervention efficacy up to 77% in suppressing dengue incidence, with efficacy rising as release coverage increased. (NIH)

One honesty note: the headline US results above are for Aedes aegypti, while Google's Florida application targets Culex quinquefasciatus (DQB strain, wAlbB). The underlying mechanism (cytoplasmic incompatibility) is species-agnostic and has been demonstrated across multiple genera, but you'd be overstating it to claim Culex-specific large-scale US efficacy data exists. Framing it as "a well-validated mechanism, already proven operationally by this applicant domestically" is accurate.

The public-health stakes are real and currently rising

• Culex quinquefasciatus is not a nuisance target. It's the primary West Nile vector in the southern US: the CDC identifies Cx. pipiens in the northern states, Cx. quinquefasciatus in the southern states, and Cx. tarsalis in the western states as the most important WNV vectors. (CDC)
• West Nile is the leading cause of arboviral disease in the United States. And it's trending the wrong way: 2025 saw 2,076 reported human WNV cases, up from 1,466 in 2024. (CDC, Vector Disease Control International)
• The disease burden is severe at the tail even though an estimated 70%–80% of WNV infections are asymptomatic and less than 1% develop neuroinvasive disease, which typically presents as encephalitis, meningitis, or acute flaccid paralysis. There's no vaccine and no specific treatment, so vector control is the only lever. (CDC, CDC)

Safety and containment — preempt the obvious objections

• Only females bite, so the releases don't add to the biting/disease burden even during the trial. The released males mate with wild females and the resulting eggs do not hatch. (NewsX)
• Wolbachia is not an engineered agent and can't infect people. It's a bacterium naturally found in at least 40 percent of all insect species. Stress that these mosquitoes are not genetically modified — this is the key distinction from the genetic-engineering approaches that draw most public alarm. You're commenting in favor of the less radical of the two technology families. (NPR)
• It's self-limiting and reversible. Because only males are released and they don't bite or persist, suppression stops when releases stop; you're not permanently re-engineering an ecosystem.
• There's regulatory precedent for the safety finding. The EPA registered Wolbachia-infected male mosquitoes (MosquitoMate's product) back in 2017, and the EPA approved the Verily project as safe for the environment. (NewsX)

Comparative argument against the status quo

The baseline isn't "no intervention," it's broadcast chemical adulticiding. A targeted, species-specific tool spares pollinators and non-target insects that pyrethroid / organophosphate fogging hits, and it sidesteps the growing insecticide-resistance problem that's eroding conventional control. If you want the comment to land with EPA reviewers specifically, the resistance-management and reduced-chemical-load angle is the one that speaks their language.

Here are Claude's thoughts on why one might oppose this:

The substantive ones, strongest first:

• Imperfect sex-sorting → population replacement. This is the most technically serious objection. Sorting males from females isn't perfect, and if fertile Wolbachia-infected females are released, the wAlbB strain can establish in the wild population. That undermines the suppression goal and isn't cleanly reversible, which cuts against the "self-limiting" selling point. It's the reason many programs bolt on irradiation (SIT) as a backstop, and a commenter can fairly ask what failsafe Google is using at this scale.
• Ecological / non-target effects. Mosquitoes are prey (birds, bats, fish, insects) and larvae process detritus. Suppression studies generally find minimal trophic disruption, but "generally minimal in prior trials" isn't the same as "established safe for this species at this scale," and there's the question of whether another vector fills the vacated niche.
• Scale and novelty. 64 million across two states is a large jump from prior US trials, on a species (Culex quinquefasciatus) that hasn't had a large US IIT deployment. A precautionary argument says the burden is on demonstrating safety at this magnitude before approving it, not after.
• Consent and who decides. Residents in release neighborhoods don't individually opt in. This was a live grievance in Fresno and the Keys, and it has an environmental-justice edge if releases concentrate in particular areas.
• Private-actor oversight. A for-profit tech company running open-environment biological releases, explicitly to support a future commercial product registration, raises accountability questions: who's liable if it goes wrong, what data is collected, who's monitoring.

Weaker but real:

• Efficacy/cost skepticism. Suppression is temporary and needs continuous re-release; mosquitoes reinvade from outside the treated zone. Critics argue the WNV burden (mostly asymptomatic, modest case counts) doesn't justify it versus conventional integrated vector management.
• Precedent, run in reverse. Your post treats normalization as a benefit. Someone can flip it: oppose precisely because it smooths the path toward gene drives, which are self-propagating and irreversible in a way Wolbachia IIT is not.