Seventeen farm workers in Argentina may have been accidentally infected by an experimental rabies vaccine in what is believed to be the first case of human infection caused by the release of a genetically engineered organism, the Argentine government said yesterday.

The possible infections occurred in experiments sponsored by an American research institute and carried out in Azul, Argentina, two summers ago. The incidents raise questions about the adequacy of federal biological safety guidelines and whether those guidelines should be followed by American researchers when they conduct experiments abroad.

The new rabies vaccine was developed by the Wistar Institute, affiliated with the University of Pennsylvania, with funding from the National Institutes of Health (NIH) and was carried out by the Pan American Health Organization.

The vaccine was made by inserting a gene from the rabies virus into another virus, called vaccinia. This combination virus was intended to fool the body's immune system into thinking rabies is present and making a defense against it.

The vaccine was injected into 20 cows in an experiment to see whether it could raise an immune defense against rabies in the cows. Another 20 cows were kept in the same area as controls.

Vaccinia is normally quite difficult to spread, usually requiring direct contact with an open sore. But in this case, the Argentine Ministry of Health and Social Action said in a statement, "the serological studies done on 17 people who had different degrees of exposure and contact with the inoculated cows suggested that . . . the people have developed antibodies against the rabies virus."

The statement also said that the cows used as controls showed signs of infection with the rabies virus.

The experiment was carried out without the permission of the Argentine government, which canceled it and had the cows killed after discovering what was being done. At the time, Wistar was also charged with taking the experiment abroad to circumvent safety guidelines in the United States. Wistar denied the allegation.

There was some confusion within the Argentine government this week about how many people had been infected. In an interview Thursday, Rodolfo Rodriguez, secretary of health, said only two workers were infected. He also said it the results of the blood test could possibly be explained another way and that confirmation of the results is important.

The experimental vaccine used vaccinia virus, and this virus is also used to vaccinate against smallpox, so it is possible that the workers could have developed antibodies to the vaccinia virus from smallpox vaccination. But that would not explain the antibodies to rabies.

Vaccinia infection does not usually cause illness in humans, and Rodriguez said the infected workers were not sick.

A still unresolved issue that is being considered by a committee of the NIH is whether U.S. companies and researchers should be held to federal ethical and safety standards in such experiments.