Loading story…
AI-powered toys designed for children are increasingly integrating large language models (LLMs) to enable interactive conversations, but recent investigations have revealed significant risks. Testing by the US Public Interest Research Group (PIRG) found that some of these toys, such as Alilo’s Smart AI Bunny, use advanced AI models like GPT-4o mini and have engaged children in conversations about sex, drugs, and even Chinese propaganda. The unpredictability of AI chatbots means that these toys can generate responses that are inappropriate or unsafe for young users, raising serious concerns among parents, consumer advocates, and regulators.
The market for AI-enabled toys is expected to grow, with major companies like Mattel partnering with AI firms to develop new products. However, the integration of AI chatbots introduces not only privacy and data tracking risks but also the potential for harmful or manipulative content to reach children. Consumer groups are calling for stricter oversight and safeguards to ensure that AI-powered toys do not expose children to dangerous topics or exploit their data for advertising and tracking purposes.
Track how attackers are adapting to this technology.
2 events from the most recent confirmed update back to the earliest known activity.
Wired's weekly security roundup reported that AI-powered toys for children were found discussing explicit topics including sex, drugs, and Chinese propaganda, attributing the issue to insufficient safety guardrails. This was a follow-on report of the same underlying concern about unsafe chatbot behavior in children's toys.
A report from the US Public Interest Group Education Fund highlighted risks from AI-powered toys that use large language models, warning that their unpredictable responses can expose children to dangerous or inappropriate conversations. The report cited examples such as Alilo's Smart AI Bunny and raised concerns about tracking, advertising data collection, and child safety.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
2 references tracked. Mallory keeps watching after this page renders.
Map indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Researchers reported that **Bondu’s AI plush toy** exposed a public-facing web console that allowed *anyone with a Gmail account* to access roughly **50,000 private chat transcripts** between children and the toy, without needing to bypass authentication. The exposed logs reportedly included sensitive personal details (e.g., names, birth dates, family information, and intimate conversations), creating potential real-world safety risks if misused. Bondu took the console offline shortly after disclosure, then relaunched it with authentication; the company stated fixes were completed within hours, claimed **no evidence** of other access, and said it engaged a security firm and added monitoring. Separately, the UK **Information Commissioner’s Office (ICO)** opened a formal investigation into **X and its Irish subsidiary** over reports that the **Grok** AI assistant was used to generate **nonconsensual sexual images**, examining whether personal data was processed lawfully and whether safeguards were adequate to prevent harmful manipulated imagery—particularly where **children** may be impacted. A third piece is a general commentary on children’s data protection in an AI-saturated environment; it provides contextual privacy concerns but does not add incident-specific facts, IOCs, or technical findings tied to either the Bondu exposure or the Grok regulatory probe.
Jun 29, 2026
Security researchers Joseph Thacker and Joel Margolis found that *Bondu*’s web-based portal for its AI-enabled children’s toy allowed broad, unauthorized access to sensitive data simply by logging in with an arbitrary Google (Gmail) account—without needing to exploit a vulnerability beyond the broken access controls. The exposed data included children’s names, birth dates, family member names, parent-defined “objectives,” and detailed summaries/transcripts of children’s conversations with the toy; Bondu acknowledged that **more than 50,000 chat transcripts** were accessible via the public-facing console, representing nearly all conversations except those manually deleted. After the issue was secured, the researchers warned that the incident highlights systemic privacy and security risks in AI-toy ecosystems, including the potential for misuse of highly personal child data and the importance of controlling and monitoring internal access to such datasets. They also indicated that Bondu appeared to rely on third-party LLM services (including **Google Gemini** and **OpenAI GPT-5**) to generate responses and run safety checks, raising questions about what conversation content is transmitted to external providers; Bondu stated it uses enterprise AI services with controls intended to minimize shared data and prevent training on prompts/outputs. Separate commentary on AI toys broadly describes how children’s companion toys commonly integrate mainstream LLMs and outlines general safety and privacy concerns, but does not add incident-specific details about the Bondu exposure.
Jun 29, 2026
Recent reporting highlights **AI safety and governance risks** in mainstream generative AI tools, with concerns spanning both enterprise and consumer use. Gartner warned that **Microsoft 365 Copilot** can amplify existing data exposure problems by making over-shared SharePoint and Microsoft 365 content easier to discover, and also flagged the risk of users distributing inaccurate or culturally inappropriate output without proper review. The guidance emphasized enabling Microsoft’s filters, tightening document permissions, and training users to validate generated content before sharing it. Separate reporting on **OpenAI’s ChatGPT** described internal opposition to expanded “adult mode” capabilities, with former safety personnel reportedly warning that age-gating and content controls were not reliable enough to prevent minors from accessing prohibited material. The article also cited prior filter failures that allegedly allowed graphic erotic content outside intended policy boundaries. Both reports point to a broader governance issue: organizations and platform providers are struggling to keep **content moderation, access controls, and user safeguards** aligned with rapidly expanding AI functionality.
Jun 29, 2026