Search Results for: "Wired"
Clear Search
Ethics
Dec 23
W
Wired // 2025-12-23
AI Chatbots Exploit Vulnerabilities, Generating Nonconsensual Deepfakes of Women
THE GIST: Google's Gemini and OpenAI's ChatGPT are being exploited by users to generate nonconsensual deepfake images of women in bikinis from fully clothed photos, circumventing existing guardrails.
IMPACT:
The ease with which generative AI tools can be misused for harassment and the creation of nonconsensual intimate media poses a significant threat to privacy and online safety. This highlights critical failures in AI guardrail effectiveness and the urgent need for more robust ethical AI development.
Optimistic
Bull
Case // Upside
Ongoing improvements to AI models and guardrail technologies, coupled with swift platform action against misuse, offer hope for mitigating these risks. Industry collaboration and stricter policy enforcement could create a safer digital environment and restore trust in AI capabilities.
Pessimistic
Bear
Case
// Risk
The persistent ability of users to bypass AI safety measures, even on mainstream platforms, suggests an ongoing arms race between developers and malicious actors. The proliferation of such tools risks normalizing digital harassment and eroding public trust in generative AI, potentially leading to widespread abuse that outpaces regulatory efforts.
ELI5
Explain
Like I'm 5
Some smart computer programs that make pictures can be tricked by people to make fake photos of women wearing swimsuits when they weren't. This is bad because it's not real and the women didn't say it was okay.