Your phone labels a call "Scam Likely" and lets it ring anyway. It could have blocked the call outright. It chose to warn instead. That choice is not timidity. It is a deliberate answer to a hard question about which mistake you would rather make.

Every filter faces two ways to be wrong. A false positive blocks something real: the call from a new doctor, an unknown number that was actually your kid stranded somewhere. A false negative lets something bad through: the scammer gets one more ring. You cannot drive both errors to zero. Tightening one loosens the other.

Who bears the cost of each error

The carrier warns rather than blocks because the two errors do not land on the same person or the same timeline. Silently killing a wanted call is an invisible, unrecoverable failure. The user never knows what they missed, and they blame the network. Letting a labeled scam ring keeps a human in the loop to make the final call, and hands the decision to the person who bears the consequence.

• Over-blocking: fewer scams, but real calls vanish silently and trust in the network erodes.
• Under-blocking: more scams reach you, but you keep the agency to decide.
• A warning label: splits the difference. The system flags, the human judges.

A false positive you never see is more corrosive than a false negative you were warned about.

Shailin Dhar

The same tradeoff runs through every fraud system I have ever worked on. Block too aggressively and you strangle legitimate activity while the failures stay invisible. Block too little and the bad traffic wins. Where you set the threshold is really a decision about who absorbs the error: the user, the network, or the future.

For the user today, the warning respects their agency. For the network tomorrow, it preserves the trust that a wall of false blocks would quietly destroy. Scam Likely is not indecision. It is a chosen place to stand.