The mood at AquaBounty Technologies a year ago was buoyant. Regulators had released a draft assessment of the company’s genetically engineered salmon, which grow faster than normal, and found them to be environmentally benign. A few months after the assessment’s comment period closed, the company began to raise more than 6,000 kilograms of salmon at its facility in Panama, in anticipation of the final approval by the US Food and Drug Administration (FDA) that would open the gates and allow the fish onto supermarket shelves.

That optimism now lies buried alongside the fish, which were culled when the approval failed to come through. The FDA says that it is still processing the more than 35,000 public comments made in response to the draft assessment. But for AquaBounty, based in Maynard, Massachusetts, this is just the latest in a series of delays spanning nearly 20 years (see Nature 497, 17–18; 2013). Many of the FDA’s deliberations have taken place behind closed doors, fuelling confusion as to the cause of the setbacks, and rumours of political interference.

As the delays have dragged on, the technology used to make AquaBounty’s salmon has become outdated. In the current excitement over targeted gene editing that allows researchers to modify individual genes without leaving traces of foreign DNA, AquaBounty’s salmon — which contain a gene from another species — seem like a relic.

But the company’s experience may hold a cautionary message. The FDA has not yet announced how it will evaluate animals engineered with gene-editing techniques. Its discussions are again occurring in private, leaving frustrated researchers to wonder whether the fruits of these technologies will meet the same fate as the beleaguered salmon. The FDA should learn from past experiences, bring these discussions before the public, and leave political considerations at the door.