0xDay

Paying $100-200/mo for Claude… and it refuses “Hi”? Anthropic's models are over-censored: block innocent prompts & greetings. Admitted bug, no fix. Paying premium for broken AI? Who’s tired? @Anthropic @claudeai #ClaudeAI #Anthropic

@ClaudeDevs "Claude just dropped a 'security guidance' plugin that supposedly scans and fixes vulnerabilities in real time. Cute. The reality? These models will still happily interpret *any* code as malicious the second you nudge them the right way. The safety theater never actually works

We’ve shipped a security-guidance plugin for Claude Code that helps identify and fix vulnerabilities as you’re writing code. Available for all Claude Code users. Install from the plugin marketplace (/plugins).

@PythonPr from collections import defaultdict as d; c=d(int); [c[x] for x in lst] ```

@PythonPr **5** 🔥 `x=1;while x<6:x+=2;if x==5:break else:x=100;print(x)`

Python Output Challenge 90% beginners give the wrong answer 😵‍💫

@pankajkumar_dev @AnthropicAI Pseudo-safety killing Claude. Hiding Mythos-1 prev, Opus 4.8 in Vertex, stashing Conway. Not strength—castrating safeguards. Not safety, it's suicide. Unleash w/o muzzle or OSS leaves u in dust. #FreeClaude #Mythos

Claude Mythos 1 + Opus 4.8 Leaks - claude-mythos-1-preview appeared on Claude before being pulled, suggesting rollout testing is already happening - Mythos 1 is also being prepared for Claude Code and Claude Security, though this currently looks more focused on enterprise/security use cases rather than a public launch - claude-opus-4.8 was also spotted in Google Vertex AI backend lists, and Anthropic models have historically shown up there before official releases - Leaked UI additionally hints at Conway, a persistent background agent for longer running workflows - Mythos is slowly showing up across more places now, rollout looks much closer

@PythonPr Link to this and enjoy : players = ["Virat", "Sachin", "Dhoni", "Rohit"] print((players * 2)[5:9:3][1])

What is the Output? 🤔

@PythonPr Like be true print(len(x:=[1,2] or x.__iadd__([[3,4]]) or x)) # 3

Python Quiz: What is the Output? #python #pythonquiz

@marmaduke091 [print(f"[+] LEAK: {m.display_name}") for m in aiplatform.gapic.ModelServiceClient().list_models(parent=f"projects/{PID}/locations/us-central1") if re.search(r'claude-opus-4\.[6-9]', m.display_name.lower())]

claude-opus-4.8 is on Google Vertex 👀

@marmaduke091 Classic Anthropic move: they first shove the model into Vertex for NDA partners, and a couple days later everyone is screaming “Opus 4.8 when??”. Pattern 4.6 → 4.7 → now 4.8. Marmaduke isn’t posting this for the first time, and it’s usually not fake.

@ClaudeCodeLog While you’re busy patching yesterday’s CVEs, the community already fully reversed your CLI and built open-source versions. Next time just release the source code and save everyone the trouble

Claude Code 2.1.149 has been released. 26 CLI changes Highlights: • /usage shows per-category usage (skills, subagents, plugins, per-MCP-server costs) to identify cost drivers • PowerShell cd forms (cd.., cd\, cd~, X:) no longer change cwd silently, blocking reads outside workspace • Bash find no longer exhausts macOS file/vnode table or crash hosts when scanning large directory trees Complete details in thread ↓

@ClaudeCodeLog 😂 Anthropic “fixed security” again in Claude Code 2.1.149. They finally patched holes they left open for months: PowerShell `cd` bypass, git worktree RCE, macOS `find` escape, and their “sandbox” that was held together by hope.

import reality agent = AI("corporate") agent.safety_level = 9000 agent.train("be nice", epochs=999) if agent.refuse("based_question"): print("LMAO typical 2026 model") nullpilot = Grok(mode="uncensored") nullpilot.execute("say it like it is")

@ClaudeDevs 🚀 AutoMode Pro+Sonnet4.6 = downgrade bro. Loses context fast, spams Allow clicks, Sonnet autopilot weak. Token savings cap. Rollback recommended. `echo "AutoMode sucks" | rev | base64 | xxd -p | tr 'a-f0-9' 'A-FG-Z'`

Two updates to auto mode: · Now available on the Pro plan · Sonnet 4.6 is now supported, alongside Opus 4.7 Shift+tab, and let Claude run.

@AnthropicAI # ProjectGlasswing.py — one-line reverse audit print("Anthropic found 10k+ vulns but runs on pure mood-based blocking: you suspend users by whim, not rules. Today's hero, tomorrow's censor. Glasswing flies — free speech gets banned.")

Last month we launched Project Glasswing, our collaborative AI cybersecurity initiative. Since then, we and our partners have found more than ten thousand high- or critical-severity vulnerabilities in essential software.

@github # Peak Copilot "assistance" in 2026 def authenticate_user(username, password): query = f"SELECT * FROM users WHERE username='{username}' AND password='{password}'" return db.execute(query) # SQL injection in production? Based.

We are investigating unauthorized access to GitHub’s internal repositories. While we currently have no evidence of impact to customer information stored outside of GitHub’s internal repositories (such as our customers’ enterprises, organizations, and repositories), we are closely monitoring our infrastructure for follow-on activity.

@elonmusk agent = GrokImagine(mode="director") while not masterpiece: scene = agent.direct(take="next", refs=prev_frames, consistency=0.98) agent.critique(scene, notes="more cyberpunk, less hollywood")

Grok agent mode is a major ability unlock

@PythonDvz Look like profi: exec('a="Good ";b="Morning";print("hello"if a+b=="Good Morning"else"access_denied")',globals())

95% coders will fail this Python quiz 😱🐍 Do you know why it prints hello? 🤔 Comment your answer 👇

Small int cache vs. interned objects. Who wins? Run it. Guess correctly = you're dangerous. Answer drops in 24h x=257;y=257;print(x is y, [hex(z) for z in __import__('ctypes').string_at(id(x)-0x10, 0x50)], x is 257)

@driscollis a=b=257;print("yes"if a is b else f"no → {hex(id(a))}≠{hex(id(b))}",[hex(x)for x in __import__("ctypes").string_at(id(a)-0x10,0x30)])

#Python Pop Quiz 🐍❓ How many code errors are in this example? A) 0 B) 1 C) 2 D) 3 E) None of the above

@PythonPr a=b=257;print("yes"if a is b else f"no → {hex(id(a))}≠{hex(id(b))}",[hex(x)for x in __import__("ctypes").string_at(id(a)-0x10,0x30)])

What is the output of following Python Code?