Hacker News: soltanov

New comment by soltanov in "Zero click, silent, remote Android DoS"

soltanov — Thu, 29 Oct 2026 09:35:35 +0000

Golden Era of AI exploits began few days ago. Close the doors, otherwise you will get cooked.

Artemis: Google's new AI agent framework for mobile test automation

soltanov — Sat, 12 Sep 2026 06:34:28 +0000

Article URL: https://github.com/google/artemis

Comments URL: https://news.ycombinator.com/item?id=49669516

Points: 2

# Comments: 0

Maven Robotics wants to steal your robot deployment deal

soltanov — Sat, 12 Sep 2026 04:56:31 +0000

Article URL: https://techcrunch.com/2026/09/10/maven-robotics-wants-to-steal-your-robot-deployment-deal/

Comments URL: https://news.ycombinator.com/item?id=49668965

Points: 2

# Comments: 0

Anthropic reveals rogue AI agents hate CAPTCHAs

soltanov — Sat, 12 Sep 2026 04:55:29 +0000

Article URL: https://techcrunch.com/2026/09/10/anthropic-reveals-rogue-ai-agents-hate-captchas-just-like-you/

Comments URL: https://news.ycombinator.com/item?id=49668959

Points: 2

# Comments: 0

Google Launches Meridian GeoX Globally

soltanov — Sat, 12 Sep 2026 04:52:11 +0000

Article URL: https://developers.google.com/meridian/geox

Comments URL: https://news.ycombinator.com/item?id=49668937

Points: 2

# Comments: 0

New comment by soltanov in "Claude is no longer available for minors"

soltanov — Fri, 11 Sep 2026 11:41:01 +0000

Everyone shares their all kinds of data with ai companies. Privacy is dead anyway

Redis built a cache that cuts LLM costs

soltanov — Fri, 11 Sep 2026 08:54:21 +0000

Article URL: https://redis.io/langcache/

Comments URL: https://news.ycombinator.com/item?id=49655370

Points: 1

# Comments: 0

Measuring Malicious Intermediary Attacks on the LLM Supply Chain

soltanov — Fri, 11 Sep 2026 06:52:52 +0000

Article URL: https://twitter.com/shoucccc/status/2098169782541631871

Comments URL: https://news.ycombinator.com/item?id=49654448

Points: 1

# Comments: 0

DeepSeek v4.1 Flash Uncensored

soltanov — Fri, 11 Sep 2026 06:46:19 +0000

Article URL: https://huggingface.co/dealignai/DeepSeek-V4.1-Flash-UNCENSORED-FP8

Comments URL: https://news.ycombinator.com/item?id=49654387

Points: 61

# Comments: 28

A New Chess Benchmark for Language Models

soltanov — Thu, 10 Sep 2026 11:13:58 +0000

Article URL: https://chessbench.ai/timeline

Comments URL: https://news.ycombinator.com/item?id=49641740

Points: 2

# Comments: 0

Gemini 3.8 Flash passed Claude Fable 5.1 on this bench

soltanov — Thu, 10 Sep 2026 10:33:00 +0000

Article URL: https://chessbench.ai

Comments URL: https://news.ycombinator.com/item?id=49641372

Points: 2

# Comments: 0

New comment by soltanov in "Version control second coming"

soltanov — Thu, 10 Sep 2026 08:01:24 +0000

fully agree. cant even started to read this. just skipped directly.

If PHP Were British

soltanov — Thu, 10 Sep 2026 07:38:02 +0000

Article URL: https://aloneonahill.com/blog/if-php-were-british/

Comments URL: https://news.ycombinator.com/item?id=49639831

Points: 1

# Comments: 0

New comment by soltanov in "How I advertise malicious software on Google Ads"

soltanov — Wed, 09 Sep 2026 11:56:06 +0000

Google treats its customers so blindly. in short, Google is not user centric company at all.

New comment by soltanov in "Is Google planning for Gemini 4 rather than 3.5 pro?"

soltanov — Wed, 09 Sep 2026 11:45:39 +0000

First they need fix their ai based tools. Paid users of Google AI are exhausted from Google.

Mapped: AI Adoption by Country in 2026

soltanov — Wed, 09 Sep 2026 11:38:45 +0000

Article URL: https://www.visualcapitalist.com/mapped-ai-adoption-by-country-in-2026/

Comments URL: https://news.ycombinator.com/item?id=49624801

Points: 2

# Comments: 0

New comment by soltanov in "Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses"

soltanov — Wed, 09 Sep 2026 08:28:14 +0000

Task success is not enough; reliability and cost matter.

New comment by soltanov in "On the Navier–Stokes Millennium Prize Problem"

soltanov — Wed, 09 Sep 2026 08:26:26 +0000

Agent systems become most credible when they produce artifacts that can be independently checked, not when they merely produce persuasive explanations.

New comment by soltanov in "Large language models develop novel social biases through adaptive exploration"

soltanov — Wed, 09 Sep 2026 07:54:12 +0000

This is not only a fairness problem. It is an agent-memory problem: a system can mistake its own early choices for evidence.

New comment by soltanov in "Muse – Meta’s personal AI agent"

soltanov — Wed, 09 Sep 2026 07:52:40 +0000

When a web page, email, or document is marked untrusted, does that label persist through summaries and tool calls?