From The Philadelphia Inquirer
(http://www.philly.com/mld/inquirer/2002/10/23/business/4348737.htm)

The program dispenses advice to children on issues such as revealing personal information.

By Anick Jesdanun
Associated Press

Wednesday October 23, 2002

NEW YORK - America Online is launching an Internet-safety campaign for children built around an automated instant-messaging buddy, or "bot," that dispenses advice in real time.

Children can add "AOLSafetyBot" to their buddy lists of friends on AOL Instant Messenger. It is programmed to answer, within seconds, such questions as whether children should agree to physical meetings with online acquaintances or reveal such personal information as their address and age.

Some experts wonder, however, whether a scripted program can always be an appropriate guide in a complicated online world, given varying age groups and parental preferences.

The SafetyBot campaign, being launched today, also includes a Web site (www.safetyclicks.com) where children can play a trivia game and watch a video featuring characters from the Cartoon Network, a unit of America Online's parent, AOL Time Warner Inc.

People who do not use AOL's instant-messaging software can also find the SafetyBot buddy on the SafetyClicks site.

Automated instant-messaging bots are not new, but past ones have been devoted mostly to marketing and promotions. Internet safety resources also exist elsewhere as Web sites - among them, Disney's SurfSwellIsland.com.

AOL said it created SafetyBot to bring safety resources to a forum that children were already familiar with. The company's instant-messaging software is the most popular on the Internet, with more than 150 million registered users.

According to the Pew Internet & American Life Project, more than 40 percent of teenagers on the Internet use instant messaging on a given day, compared with 11 percent for online adults.

"Instant messaging clearly is a form of communications that they enjoy, so there was a natural predisposition to using a bot," said Tatiana Gau, AOL's senior vice president for integrity assurance.

According to a 2000 study by the Crimes Against Children Research Center at the University of New Hampshire, one in five youths age 10 to 17 received unwanted sexual solicitations over the Internet during the previous year. Only a quarter of them told a parent.

Will children use the bot instead of asking parents for advice? Will parents depend on the bot to supervise their children?

Gau said the bot was not meant as a substitute for other safety resources.

"Obviously, as a bot, it has intelligence and the ability to answer and handle certain questions," she said. But the bot has "certain limitations, as all bots do."

Many of the answers emphasize telling parents if, say, an online acquaintance asks for a meeting or personal information.

Parry Aftab, a leading Internet-safety expert, applauded efforts to make learning about safety fun.

"The kids will play with it, and if they play with it, maybe they will learn something," she said.

But she cautioned that correct answers may depend on age and other factors - for instance, some parents may want to handle meeting strangers differently. The AOL bot offers only generic responses.

And even safety experts may disagree on the proper approach. Aftab said she used to recommend that children give strangers a false name - until someone pointed out that children might then consider lying permissible behavior.

"There is not always one clear answer," she said.

The AOL bot, made available to a reporter for testing, was good about giving relevant answers on general safety issues, such as sending photos to strangers, protecting passwords, and confronting bad language.

But it did not always answer questions head-on.

For example, the question "Could you meet me at Kmart?" returned a warning never to meet online friends in person without a parent. The question "Could you come to Kmart with me?" returned a generic message introducing the bot.

Off-topic questions occasionally yielded humorous answers.

What are your hobbies? "I like to dance. (The 'electric slide' is my favorite.) I also like to read and surf the Internet."

Why are you annoying? "Well, I am a bot."

Other times, even with questions related to Internet safety, the bot said it could not understand and directed the user to ask again, visit a menu of safety tips, or ask a parent.

AOL officials say the bot, which only responds when addressed, was programmed to make such replies rather than guess and potentially give a wrong answer. More questions and answers will be added over time.