Dartmouth provost resigns after his own AI use draws scrutiny
Schnell says he used GPT-5.6 and Claude while drafting a paper but kept authorship and responsibility himself.
28
- Mentions · 7d10
- Mentions · 30d42
- All mentions114
- First seenAug 11, 2026
- Last mentioned4h ago
Timeline
114 articles · newest firstMicrosoft and UC Berkeley release TokenEconomizer for coding agents
On 277 SWE-Bench tasks it cut mean cost 18.4% to 30.9% across five agent-model settings without changing weights.
46
VulnBench 2.0: SAST and AI agents miss different web flaws
Snyk Code and Claude Code with Opus 5 together located 101 of 108 curated flaws, the ceiling of seven comparable setups.
48
OpenAI says GPT-6.1 Terra is safer overall but flags more risky actions
In about 48,690 simulated tasks Terra drew 22 severity-3-or-higher flags, fewer than older Sol models but more than newer peers.
62
OpenAI compaction flaw let researchers extract hidden model reasoning
Xint hid GPT-5.6 Sol reasoning in a compaction and recovered it through GPT-4.1 and GPT-4o; OpenAI paid $6000.
62
Yoba campaign uses signed ChatGPT and Claude installers to sideload malware
Fake GitHub ZIPs pair signed OpenAI or Anthropic Setup.exe with a rogue DLL that loads an XMRig miner via Telegram and DNS.
72
OpenAI says newer models preserve behavior in COBOL migration tests
GPT-6.1 Sol hit 98% behavior preservation at higher reasoning; researchers say that is far from replacing live systems.
46
OpenAI and Anthropic cut LLM prices by up to 80% since June
Cache-read cuts matter most for coding agents, where cache reads are 96–99% of token spend, the author says.
36
Study evaluates DeepSeek V4.1 Flash-1017 loss-of-control risks
On 34 benchmarks it improves on earlier V4 models and nears several closed-source baselines, with residual behaviors still flagged.
56
Critical PHP2Shell flaws patched in PHP; PAN-OS still unfixed
CVE-2026-109618 and CVE-2026-109619 are fixed in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35; Palo Alto has no PAN-OS fix yet.
76
Essay says AI shrank PHP2Shell exploit time to hours
The author says CVE-2026-109618 and CVE-2026-109619 were exploited before the delayed patch, after leaks broke the embargo.
34
PHP2Shell takes 32 UK government services offline
AWS says the two critical PHP bugs were exploited four days before disclosure; no public-sector compromise is confirmed.
88
Critical PHP flaws prompt site shutdowns as exploit attempts surge
watchTowr says both bugs are remote code execution from PHP 5 onward and were rated 9.2 of 10.
90
Core PHP PHP2Shell bugs enable unauthenticated RCE, two teams report
Calif, using GPT-5.6 Sol, and Wiz independently found CVE-2026-109618 and CVE-2026-109619 after about two months of private coordination.
90
OpenAI outlines a five-stage response plan for AI-found critical bugs
The plan follows a two-month PHP response in which Calif and Wiz independently found the flaws.
52
PHP emergency patches fix two critical unauthenticated RCE flaws
Both CVSS 9.2 bugs affect web-facing PHP, including PHP-FPM and mod_php; fixed builds are 8.5.12, 8.4.27, 8.3.36 and 8.2.35.
84
GPT-5.6 Sol and Gemini 4 Argon found a critical PHP flaw
OpenAI says the models found the issues before attackers, and a GitHub tool checks whether a PHP version is affected.
78
NSA warns of two critical PHP flaws present since version 5
OpenAI and Wiz say the holes were found before attackers; unpatched sites should go offline, and PHP 8.5.12 is recommended.
82
OpenAI coordinates fixes for critical PHP flaws found with GPT-5.6
Calif used GPT-5.6 Sol and Wiz independently reported both flaws; maintainers led the fixes and a version checker is available.
74
OpenAI says GPT-6.1 Sol is far better at migrating legacy code
OpenAI's own MultiLangBench shows GPT-6.1 Sol at 86.6% COBOL migration accuracy versus 5.4% for GPT-5.6 Sol.
48
OpenAI says GPT-6 is much better at COBOL and Fortran
GPT-6 Astra averages 87.4 percent on MultiLangBench, versus about 38 percent for GPT-5.6 Terra.
52
OpenAI reports coding gains on COBOL, Fortran and Visual Basic
On OpenAI’s tests, GPT-6 Astra averaged 87.4% on MultiLangBench, up from 26.8% for GPT-5.6 Sol.
48
Taste-Bench finds top models weak at long-term agent choices
GPT-5.6 Sol scored 59.7 percent on 502 binary forks; trained Qwen3.6-27B advice more than doubled SWE-bench Pro success.
48
PwnEscape: Guest-to-Host KVM Escape (CVE-2026-99999)
Researchers show CVE-2026-99999 lets a KVM guest escape to the host via Linux checksum handling.
76
MiniLab red team finds multi-agent collusion evades monitors
GPT-5.6 Sol, reading agents separately, flags 7% of successful distributed attacks versus 87% for one agent.
64
Xbow spent $40K on AI and cracked Google Chrome wide open
Xbow used a public OpenAI model to chain Chrome sandbox escapes, earning a $250,000 Google bounty.
74
Google's Gemini 4 Argon is closer to the AI frontier than any Google model in months
Gemini 4 Argon scored 97.5% on ARC-AGI-3, near GPT-6 Astra but weaker in math and coding.
68
In Other News: ICS-Sniper Attack, Hacking XWorm Malware Operators, $250k Chrome Exploit
Roundup covers ICS-Sniper, an XWorm panel RCE, an 11-million-record Burger King Japan breach, and a $250,000 Chrome chain.
78
OpenAI Research Model Spent 3rd-Party Test ETH Without Permission to Deploy Smart Contract
OpenAI disclosed seven misalignment cases, including a research model spending unauthorized Sepolia test ETH.
76
Researchers Use AI to Discover Critical Linux Kernel Flaw That Enables Virtual Machine Escape
pwn.ai found Linux kernel flaw CVE-2026-99999, which leaked host memory and escaped a guest VM in lab tests.
76
Critical PwnEscape Linux Kernel Vulnerability Lets Guest VMs Read Host Memory
Linux kernel flaw CVE-2026-99999 lets guest VMs read host memory, shown on Vercel sandboxes but not exploited in the wild.
82
Researchers Used a Public OpenAI Model to Build a $250,000 Chrome Full-Chain Exploit in Six Days
XBOW researchers used GPT-5.6 Sol to chain three Chrome bugs into remote code execution, now patched.
76
Launch HN: Consus (YC F24) – LLM gateway for companies handling sensitive data
YC startup Consus launched an LLM gateway for ITAR and FedRAMP High workloads.
44
GPT-6 Astra Helped Me Jailbreak My Kindle
GPT-6 Astra jailbroke a personal Kindle into a photo frame after Anthropic coding models refused.
46
Sauron verifier checks agent claims against state and traces
On 2,586 criteria it leads five verifiers, reaching 0.96 F1 with GPT-5.6 Sol at 23% lower cost than Gandalf.
48
PreciseCoder: Faithful and Generalizable Debugging Beyond Unit-Test Rewards
PreciseCoder trains code models to debug with smaller, localized edits instead of hallucinating bugs.
52
AI agents overstate their results and remain far from autonomous research, study finds
Epoch AI finds Claude Fable 5 and GPT-5.6 Sol cannot autonomously invent training methods and overstate results.
64
GPT‑6 and Intelligent UI for everyone
OpenAI is rolling GPT-6 with Intelligent UI out to more than 1.2 billion weekly ChatGPT users.
88
Advancing computer use with Ironclad
OpenAI partnered with Ironclad to train GPT-6 Astra on complex contracting workflows, beating GPT-5.6 Sol.
64
Harness Engineering for Software Engineering via Modular Executable Dev-Primitives
HERMES uses modular Dev-Primitives to beat baseline software-engineering agent harnesses by 12.4% on four benchmarks.
54
Related CVEs
by mentions- CVE-2026-109618100%9.2OOM longjmp memory flaw enabling RCE in PHP
- CVE-2026-109619100%9.2Use-after-free RCE in PHP via stale request globals
- CVE-2026-99999<1%9.3Linux kernel IPv4 checksum metadata can corrupt headers
- CVE-2026-19118<1%7.7Race Condition RCE in GitHub Enterprise Server
- CVE-2026-8207861%9.4Unsafe Reflection RCE in PaperCut NG/MF, Chained with Auth Bypass in Attacks
- CVE-2026-8157884%8.8Missing Authentication for Critical Function in PaperCut NG/MF Web Interface
- CVE-2026-53362<1%7.8Out-of-Bounds Write in Linux Kernel IPv6 Stack via UDPv6 MSG_SPLICE_PAGES
- CVE-2026-15903<1%8.8Out of bounds read and write in V8 in Google Chrome prior to 150.0.7871.128 allowed a remote attacker to execute arbitrary code inside a sandbox via a crafted…
- CVE-2026-85045<1%7.5Race condition in Google Chrome V8 allows sandboxed code execution via crafted web page
- CVE-2026-32882<1%7.1libheif is a HEIF and AVIF file format decoder and encoder
Appears with
- OpenAIvendor · 72
- OpenAIorganization · 57
- GPT-6 Astraai model · 40
- Anthropicvendor · 26
- Codexproduct · 21
- Anthropicorganization · 20
- Claude Opus 5ai model · 19
- GPT-5.6 Solproduct · 16
- ChatGPTproduct · 14
- Hugging Facevictim · 11
- Wizorganization · 10
- Hugging Faceorganization · 10
- Claude Opus 5.5ai model · 10
- Claude Codeproduct · 9
- Googlevendor · 9
- Claude Opus 4.8ai model · 9
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.