"Casting such a wide regulatory net runs the very real risk of cutting us off from the many -- and potentially life-saving -- benefits these technologies offer."
The comments come amid intense debate over deployment of facial recognition for applications in security and law enforcement, among others.
San Francisco and other cities have moved to ban facial recognition by government entities and privacy activists have called for better guarantees against errors and bias.
Amazon recently said it supported regulations for facial recognition. Microsoft last year announced it was adopting a set of principles for the technology, and called for new laws to avoid a "dystopian" future.
IBM said that instead of banning all facial recognition, policymakers should employ "precision regulation" in cases where there is "greater risk of societal harm."
IBM weighed in Tuesday on the policy debate over facial recognition technology, arguing against an outright ban but calling for "precision regulation" to protect privacy and civil liberties.
In a white paper posted on its website, the US computing giant said policymakers should understand that "not all technology lumped under the umbrella of 'facial recognition' is the same."
IBM said uneasiness about artificial intelligence technology which can use face scans for identification was reasonable.
"However, blanket bans on technology are not the answer to concerns around specific use cases," said the paper by IBM chief privacy officer Christina Montgomery and Ryan Hagemann, co-director the IBM policy lab.
The company said a full ban might deny consumers the convenience of less frustrating air travel or prevent first responders rapidly identifying natural disaster victims.
But IBM also said there were uses that should remain off-limits, such as mass surveillance or racial profiling.
"Providers of facial recognition technology must be accountable for ensuring they don't facilitate human rights abuses by deploying technologies such as facial matching in regimes known for human rights violations," said Montgomery and Hagemann.
IBM said any rules should be based on "notice and consent" when facial recognition is used to verify someone's identity.
That would mean stores seeking to "customize" someone's experience being required to provide clear notification that face analytics were being used.