I spent $266 and four AI models to own my tablet. GLM-5.3 finished it in a day

	I spent $266 and four AI models to own my tablet. GLM-5.3 finished it in a day(ericpardee.github.io)
	684 points by dr_pardee 18 days ago \| 289 comments
	tl;dr: After Amazon's software repeatedly shut down his Fire HD 10 kiosk tablet, the author spent five months and $266 using multiple LLMs to root it. Claude and OpenAI's models refused on safety grounds, but Chinese models Kimi K3 and GLM-5.2/5.3 identified an unpatched 2022 Mali GPU CVE, built a working exploit, and ultimately removed 100 Amazon packages—with GLM-5.3 finishing the job in a single day. The piece argues that frontier security capability is now rentable by anyone with a credit card and patience, while highlighting the widening gap between US and Chinese models on cybersecurity assistance.
	HN Discussion: ↑AI agents amplify existing expertise rather than replace it, validating the article's premise about rentable capability ↑Unleashing AI models to reverse engineer hardware for open-source support is a promising future direction ~The writeup has an off-putting AI-generated tone but the core finding about model capabilities is valid ~Chinese models miss obvious failure points that Claude/Codex catch, complicating the article's framing of Chinese model superiority ↓Without source code and reproducibility, the exploit claims are hard to verify