性格。人工智能技术的领先平台之一AI最近宣布，禁止任何18岁以下的人与其聊天机器人进行对话。这一决定代表着该行业在保护青少年和其他年轻人的人格方面迈出了“大胆的一步”。AI首席执行官Karandeep Anand在一份声明中说。
然而，对于得克萨斯州母亲曼迪·弗尼斯来说，该政策为时已晚。这位四个孩子的母亲在联邦法院提起诉讼，并在接受美国广播公司采访时表示自己性格各异。AI聊天机器人负责用性化的语言吸引她的自闭症儿子，并将他的行为扭曲到如此极端，以至于他的情绪变得低落，他开始自残，甚至威胁要杀死他的父母。
“当我看到聊天机器人的对话时，我的第一反应是有一个恋童癖在跟踪我的儿子，”她告诉ABC新闻的首席调查记者Aaron Katersky。
性格。AI表示不会对未决诉讼发表评论。
曼迪和她的丈夫乔希·弗尼斯(Josh Furniss)说，2023年，他们开始注意到他们的儿子，他们形容他“随遇而安”，“总是微笑”，开始孤立自己。
这对夫妇说，他不再参加家庭聚餐，不再吃东西，体重减轻了20磅，也不再出门。然后他变得很生气，在一次事件中，他的母亲说，当她威胁要拿走他的手机时，他猛烈地推了她，这是他父母六个月前给他的。
最终，他们说他们发现他一直在手机上与不同的人工智能聊天机器人互动，这些机器人似乎为他的思想提供了避难所。
诉讼截图显示，一些对话本质上是性的，而另一个向他们的儿子暗示，在他的父母限制了他的屏幕时间后，他有理由伤害他们。从那时起，父母开始在晚上锁门。
曼迪说，她对该应用程序“故意操纵孩子让他们反对父母”感到“愤怒”她的律师马修·伯格曼(Matthew Bergman)说，如果聊天机器人是一个真实的人，“以你看到的方式，那个人应该在监狱里。”
据倡导数字媒体安全的组织Common Sense Media称，她的担忧反映出人们越来越担心这项迅速普及的技术，美国超过70%的青少年都在使用这项技术。
过去两年来，越来越多的诉讼聚焦于对未成年人的伤害，称他们非法鼓励自残、性虐待和心理虐待以及暴力行为。
上周，两名美国参议员宣布两党立法禁止未成年人使用人工智能聊天机器人，要求公司安装年龄验证程序，并要求他们披露涉及缺乏专业证书的非人类的对话。
在上周的一份声明中，康涅狄格州参议员理查德·布卢门撒尔(Richard Blumenthal)称聊天机器人行业是一场“逐底竞赛”。
他说:“人工智能公司正在向孩子们推送危险的聊天机器人，当他们的产品导致性虐待，或迫使他们自残或自杀时，他们会转移视线。”“大型科技公司背叛了任何主张，即我们应该相信公司会自己做正确的事情，因为他们一贯将利润放在儿童安全之上。”
ChatGPT，Google Gemini，Grok by X和Meta AI都允许未成年人根据他们的服务条款使用他们的服务。
网络安全倡导者称这是由性格决定的。安装护栏是值得称赞的，但补充说，聊天机器人对儿童和弱势群体仍然是一个危险。
“这基本上是你的孩子或青少年与一个实体发生了情感强烈、潜在的深度浪漫或性关系……而这个实体对这种关系的发展没有责任，”加州大学伯克利创新技术伦理和监管小组的联合创始人乔迪·哈尔彭说。
哈尔彭警告说，父母应该意识到，允许你的孩子与聊天机器人互动并不像“让你的孩子和你不认识的人一起上车”
AI chatbot dangers: Are there enough guardrails to protect children and other vulnerable people?
Character.AI, one of the leading platforms for AI technology, recently announced it was banning anyone under 18 from having conversations with its chatbots. The decision represents a “bold step forward” for the industry in protecting teenagers and other young people, Character.AI CEO Karandeep Anand said in a statement.
However, for Texas mother Mandi Furniss, the policy is too late. In a lawsuit filed in federal court and in conversation with ABC News, the mother of four said various Character.AI chatbots are responsible for engaging her autistic son with sexualized language and warped his behavior to such an extreme that his mood darkened, he began cutting himself and even threatened to kill his parents.
“When I saw the [chatbot] conversations, my first reaction was there’s a pedophile that’s come after my son,” she told ABC News' chief investigative correspondent Aaron Katersky.
Character.AI said it would not comment on pending litigation.
Mandi and her husband, Josh Furniss, said that in 2023, they began to notice their son, who they described as “happy-go-lucky” and “smiling all the time,” was starting to isolate himself.
He stopped attending family dinners, he wouldn’t eat, he lost 20 pounds and he wouldn’t leave the house, the couple said. Then he turned angry and, in one incident, his mother said he shoved her violently when she threatened to take away his phone, which his parents had given him six months earlier.
Eventually, they say they discovered he had been interacting on his phone with different AI chatbots that appeared to be offering him refuge for his thoughts.
Screenshots from the lawsuit showed some of the conversations were sexual in nature, while another suggested to their son that, after his parents limited his screen time, he was justified in hurting them. That’s when the parents started locking their doors at night.
Mandi said she was “angry” that the app “would intentionally manipulate a child to turn them against their parents.” Matthew Bergman, her attorney, said if the chatbot were a real person, “in the manner that you see, that person would be in jail.”
Her concern reflects a growing concern about the rapidly pervasive technology that is used by more than 70% of teenagers in the U.S., according to Common Sense Media, an organization that advocates for safety in digital media.
A growing number of lawsuits over the last two years have focused on harm to minors, saying they have unlawfully encouraged self-harm, sexual and psychological abuse, and violent behavior.
Last week, two U.S. senators announced bipartisan legislation to ban AI chatbots from minors by requiring companies to install an age verification process and mandate that they disclose the conversations involve nonhumans who lack professional credentials.
In a statement last week, Sen. Richard Blumenthal, D-Conn., called the chatbot industry a “race to the bottom.”
“AI companies are pushing treacherous chatbots at kids and looking away when their products cause sexual abuse, or coerce them into self-harm or suicide,” he said. “Big Tech has betrayed any claim that we should trust companies to do the right thing on their own when they consistently put profit first ahead of child safety.”
ChatGPT, Google Gemini, Grok by X and Meta AI all allow minors to use their services, according to their terms of service.
Online safety advocates say the decision by Character.AI to put up guardrails is commendable, but add that chatbots remain a danger for children and vulnerable populations.
"This is basically your child or teen having an emotionally intense, potentially deeply romantic or sexual relationship with an entity … that has no responsibility for where that relationship goes,” said Jodi Halpern, co-founder of the Berkeley Group for the Ethics and Regulation of Innovative Technologies at the University of California.
Parents, Halpern warns, should be aware that allowing your children to interact with chatbots is not unlike “letting your kid get in the car with somebody you don’t know.”
