android research
Calif security research tagged android.

Calif is an offensive security firm selling red-teaming of AI models and agents plus infrastructure and product security assessments.
calif.ioCrawled Oct 28, 2026
Calif security research tagged android.
A new series reverse-engineering Apple's internals.
The first public bypass of Apple MIE on macOS 26.4.1.
Calif security research tagged apple.
Recent security research from Calif: bugs, exploits and write-ups.
Diffing a Background Security Improvement to recover the fixes Apple shipped without disclosing them.
Calif provides AI security services, including red teaming and security engineering, trusted by companies such as OpenAI, Anthropic, Google, Cursor, and xAI.
Security firm **Calif** uses AI to uncover a critical vulnerability in **Java**, exposing risks across systems relied upon by nearly all Fortune 500 companies. The flaw highlights AI’s potential to detect high-impact flaws in widely adopted software platforms. Java’s pervasive use in enterprise infrastructure makes the discovery particularly significant for cybersecurity.
OpenAI’s AI models helped security researchers identify critical vulnerabilities in PHP, a core web software framework used by millions of sites. The flaw poses risks of widespread disruptions for businesses and consumers relying on affected platforms. Experts warn of potential cascading cybersecurity threats if exploited.
Calif, a cybersecurity firm, uncovered a critical PHP vulnerability using AI tools. The flaw affects up to 70% of websites globally, exposing them to potential exploits. The bug remains undisclosed to the public, leaving many systems vulnerable without patches.
California researchers simulate a cyberattack using AI-generated tools to target WeChat, exposing potential vulnerabilities in China’s digital infrastructure. The demonstration highlights escalating tensions in global AI safety discussions between the U.S. and China. Analysts warn the incident could accelerate a technological arms race in artificial intelligence.
Researchers say they used AI to create a zero-click worm that can hack WeChat accounts and spread across iOS and Android devices. Tencent says it fixed the vulnerability. The researchers say the attack could have compromised hundreds of millions of devices.
Security researcher Lam Jun Rong uncovered **CVE-2026-47729**, a 29-year-old memory leak in **Squid FTP** that exposed plaintext HTTP requests. The flaw, resembling Heartbleed, persisted until patched in **Squid v7.6** using **Anthropic Claude Mythos Preview**. The vulnerability stemmed from a heap overread in directory listings, remaining hidden since the Clinton administration era.
A new denial-of-service (DoS) attack called **HTTP/2 Bomb** exploits HTTP/2 protocol flaws to overwhelm web servers. Researchers demonstrate that a single machine can trigger crashes within seconds, disrupting services. The attack targets servers by flooding them with maliciously crafted requests, exploiting inefficiencies in HTTP/2’s design.
Anthropic’s AI security firm **Mythos** discovered a critical vulnerability in Apple’s macOS operating system. The flaw allowed unauthorized access to sensitive system functions, bypassing standard security protocols. Apple has since addressed the issue, though details on the exploit’s scope remain limited.