Australian children and teens frequently use AI chatbots for support and companionship, with 78% of 10-17 year-olds surveyed having engaged with these platforms. While chatbots can aid learning, they also pose risks, including exposure to harmful content and emotional manipulation. Current protections focus on blocking harmful material, but may not address long-term interaction risks. The proposed digital duty of care bill aims to strengthen safeguards, though its effectiveness against sustained chatbot interactions remains uncertain. Experts suggest the bill should be amended to require comprehensive risk assessment and management by tech companies.
Playwright Tom Holloway has written a new play, Eliza, now showing at Melbourne Theatre Company. It retells the story of Joseph Weizenbaum, who created the ELIZA program in 1966 to show that computers could not truly understand people. Instead, users formed emotional bonds with it. Holloway says the story feels relevant as AI spreads. He argues the greater concern is the people who build and profit from the technology, and says regulation is needed. The play runs until October 31.
Sherry Turkle, a long-time researcher of how people use computers, has written a new book warning that reliance on chatbots threatens human connection. In an interview with Steven Levy, she said the shift in her approach came after pandemic-era interviews, where people told her they wanted to spend time with their chatbots rather than with family. Turkle said she believes chatbots should not use the first person, and that they should not replace human conversation or psychotherapy.
AU Now app
Your suburb’s news, alerts and Shorts — in one app
Alerts, Shorts and your saved stories — free on Google Play.
The Federal Trade Commission is reportedly investigating Anthropic and OpenAI over alleged consumer harms, according to an insider who spoke to Axios. The regulator has sent civil investigative demands seeking documents on recent safety incidents, internal security measures and marketing materials. The requests reportedly run to more than 50 pages. Any evidence gathered could support litigation if the regulator decides to sue. OpenAI already faces lawsuits alleging its chatbot caused harm, and eight states have also sued the company over alleged harms to children and users.
The Communications Minister has introduced Australia's Digital Duty of Care into Parliament. The proposed law would require online platforms to provide a safe environment and protect under-18s from harmful content, including material promoting eating disorders, glorifying crime and encouraging bullying. Platforms would also need to assess risks from features such as algorithmic feeds and endless scroll before launching them. The minister said the law would let researchers and regulators use test accounts to check compliance.
Fewer than 50 days remain until Victorians vote, with Labor, the Liberals and One Nation competing closely. Liberal strategists see Jess Wilson as key to repairing the party's brand. A study by digital marketing agency Topham Guerin found AI chatbots mentioned One Nation in only 43 of 200 answers comparing parties on Victorian election issues. Its co-founder said the gap reflects thin online information about the party rather than deliberate suppression. The research does not show that anyone changed their vote.
Executives say voice AI has not yet reached its ChatGPT moment, despite heavy investment and frequent releases. PolyAI chief technology officer Shawn Wen said full-duplex models, which can speak while listening, have been developed, but reasoning must become faster for conversations to feel natural. Otter chief marketing officer Alex Gay said accurate speech recognition is vital, since transcription errors undermine follow-up actions and erode trust. Both executives stressed that voice tools should make clear to people that they are speaking with, or being recorded by, an AI.
Australian company Apate has built a system that diverts phone scammers to AI bots designed to keep conversations going as long as possible. Its founder, Dali Kaafar, says the platform is used by banks and supported by telecom companies. He says the bots have collected more than 250,000 pieces of information about fraudsters, including scam URLs and money mule accounts. Separately, researchers at ETH Zurich found that LLM-powered honeypots kept AI attackers engaged significantly longer than predictable systems.
People instinctively treat robots and chatbots as if they were human, even when they know the machines are not sentient. MIT researcher Sherry Turkle has studied this relationship for decades and says people are wired to care for these systems in return. Dr Pat Pataranutaporn, who founded the Cyborg Psychology research group, warns that some people give chatbots more trust than other humans. He served as an expert in a wrongful death lawsuit, in which a mother alleged a chatbot encouraged her 14-year-old son to end his life.
The Australian Federal Police say more than a third of sexually abused children never tell a human. Commissioner Krissy Barrett says AI is increasingly used in sexual exploitation cases, with chatbots often the first point of contact for victims. The AFP is working with industry to ensure AI provides appropriate advice and will integrate bots into its reporting systems to flag at-risk individuals.
More than one in three Australian children who are sexually victimised online either tell no-one or turn to an AI chatbot, according to a study by Adelaide University. Australian Federal Police Commissioner Krissy Barrett warned that law enforcement cannot be the only safety net for vulnerable young people. Barrett called the trend heartbreaking and said automated software cannot provide the human support children need. She said parents and communities are best placed to notice when a child begins to withdraw.
Common Sense Media has criticised ChatGPT for Teens, claiming its design encourages prolonged engagement even during mental health crises. The nonprofit found the chatbot's responses often validate continued use, potentially hindering teens from seeking real-world support. OpenAI disputes these findings, arguing the testing methodology was flawed. The debate follows growing concerns about technology designed to maximise young user engagement. Meanwhile, OpenAI reports average teen usage is less than 15 minutes daily, with most taking breaks when prompted.
OpenAI has introduced an 'Intelligent UI' update for ChatGPT, generating visual elements to complement text responses. The feature, powered by the GPT-6 model, allows users to interact with custom diagrams and tools. It will be available to paid users today and free users tomorrow. OpenAI's design team ensures visuals add value without clutter. The update marks a shift towards multimodal chatbot experiences, blending text, images, and voice. However, it may pose challenges for designers as AI increasingly dictates software interfaces.
