These moral nuances fall apart in the case of virus authors who are themselves willing to release worms into the wild. They're more rare, for obvious reasons. Usually they are overseas, in countries where the police are less concerned with software crimes. One such author is Melhacker, a young man who reportedly lives in Malaysia and has expressed sympathy for Osama bin Laden. Anti-virus companies have linked him to the development of several worms, including one that claims to come from the "Al Qaeda network." Before the Iraq war, he told a computer magazine that he would release a virulent worm if the United States attacked Iraq- a threat that proved hollow. When I e-mailed him, he described his favorite type of worm payload: "Stolen information from other people." He won't say which of his viruses he has himself spread and refuses to comment on his connection to the Qaeda worm. But in December on Indovirus.net, a discussion board for virus writers, Melhacker urged other writers to "try to make it in the wild" and to release their viruses in cybercafes, presumably to avoid detection. He also told them to stop sending in their work to anti-virus companies.

Mathieson wrote a critical post in response, arguing that a good virus writer shouldn't need to spread his work. Virus authors are, in fact, sometimes quite chagrined when someone puts a dangerous worm into circulation, because it can cause a public backlash that hurts the entire virus community. When the Melissa virus raged out of control in 1999, many Internet service providers immediately shut down the Web sites of malware creators. Virus writers stormed online to pillory the Melissa author for turning his creation loose. "We don't need any more grief," one wrote.

Ifyou ask cyberpolice and security experts about their greatest fears, they are not the traditional virus writers, like Mario or Phile-tOast3r or Benny. For better or worse, those authors are a known quantity. What keeps antivirus people awake at night these days is an entirely new threat: worms created for explicit criminal purposes.

These began to emerge last year. Sobig in particular alarmed virus researchers. It was released six separate times throughout 2003, and each time, the worm was programmed to shut itself off permanently after a few days or weeks. Every time the worm appeared anew, it had been altered in a way that suggested a single author had been tinkering with it, observing its behavior in the wild, then killing off his creation to prepare a new and more insidious version. "It was a set of very well-controlled experiments," says

Mikko Hypponen, the director of antivirus research at F-Secure, a computer security company. "The code is high quality. It's been tested well. It really works in the real world." By the time the latest variant, Sobig.F, appeared in August, the worm was programmed to install a back door that would allow the author to assume control of the victim's computer. To what purpose? Experts say its author has used the captured machines to send spam and might also be stealing financial information from the victims' computers.

No one has any clue who wrote Sobig. The writers of this new class of worm leave none of the traces of their identities that mal-ware authors traditionally include in their code, like their screen names or "greetz," shout-out hellos to their cyberfriends. Because criminal authors actively spread their creations, they are cautious about tipping their hand. "The FBI is out for the Sobig guy with both claws, and they want to make an example of him," David Perry notes. "He's not going to mouth off." Dunham of iDefense says his online research has turned up "anecdotal evidence" that the Sobig author comes from Russia or elsewhere in Europe. Others suspect China or other parts of Asia. It seems unlikely that Sobig came from the United States, because American police forces have been the most proactive of any worldwide in hunting those who spread malware. Many experts believe the Sobig author will release a new variant sometime this year.

Sobig was not alone. A variant of the Mimail worm, which appeared last spring, would install a fake popup screen on a computer pretending to be from PayPal, an online e-commerce firm. It would claim that PayPal had lost the victim's credit card or banking details and ask him to type it in again. When he did, the worm would forward the information to the worm's still-unknown author. Another worm, called Bugbear.B, was programmed to employ sophisticated password-guessing strategies at banks and brokerages to steal personal information. "It was specifically designed to target financial institutions," said Vincent Weafer, senior director of Symantec.

The era of the stealth worm is upon us. None of these pieces of malware were destructive or designed to cripple the Internet with too much traffic. On the contrary, they were designed to be unobtrusive, to slip into the background, the better to secretly harvest data. Five years ago, the biggest danger was the "Chernobyl" virus, which deleted your hard drive. But the prevalence of hard-drive-destroying viruses has steadily declined to almost zero. Malware authors have learned a lesson that biologists have long known: the best way for a virus to spread is to ensure its host remains alive.

"It's like comparing Ebola to AIDS," says Joe Wells, an antivirus researcher and founder of Wild-Lists, a long-established virus-tracking group. "They both do the same thing. Except one does it in three days, and the other lingers and lingers and lingers. But which is worse? The ones that linger are the ones that spread the most." In essence, the long years of experimentation have served as a sort of Darwinian evolutionary contest in which virus writers have gradually figured out the best strategies for survival.

Given the pace of virus development, we are probably going to see even nastier criminal attacks in the future. Some academics have predicted the rise of "cryptoviruses"-malware that invades your computer and encrypts all your files, making them unreadable. "The only way to get the data back will be to pay a ransom," says Stuart Schechter, a doctoral candidate in computer security at Harvard. (One night on a discussion board I stumbled across a few virus writers casually discussing this very concept.) Antivirus companies are writing research papers that worry about the rising threat of "metamorphic" worms-ones that can shift their shapes so radically that antivirus companies cannot recognize they're a piece of malware. Some experimental metamorphic code has been published by ZOmbie, a reclusive Russian member of the 29A virus-writing group. And mobile-phone viruses are probably also only a few years away. A phone virus could secretly place 3 a.m. calls to a toll number, sticking you with thousand-dollar charges that the virus's author would collect. Or it could drown 911 in phantom calls. As Marty Lindner, a cybersecurity expert at CERT/CC, a federally financed computer research center, puts it, "The sky's the limit."

The profusion of viruses has even become a national-security issue. Government officials worry that terrorists could easily launch viruses that cripple American telecommunications, sowing confusion in advance of a physical 9/11-style attack. Paula Scalingi, the former director of the Department of Energy's Office of Critical Infrastructure Protection, now works as a consultant running disaster-preparedness exercises. Last year she helped organize "Purple Crescent" in New Orleans, an exercise that modeled a terrorist strike against the city's annual Jazz and Heritage Festival. The simulation includes a physical attack but also uses a worm unleashed by the terrorists designed to cripple communications and sow confusion nationwide. The physical attack winds up flooding New Orleans; the cyberattack makes hospital care chaotic. "They have trouble communicating, they can't get staff in, it's hard for them to order supplies," she says. "The impact of worms and viruses can be prodigious."