soltanov

3 ms·

31.
Benchmark for AI Assistants(assistantbenchmark.com)
1 points by soltanov 2mo ago|0 comments
32.
by soltanov 2mo ago
That is why we need true open source ai models. Otherwise closed or open weight doesnt matter, all of them read, review, use or sell our datas.
33.
by soltanov 2mo ago
https://x.com/visegrad24/status/2099467072547791057
34.
The Revolut breach just escalated(twitter.com)
4 points by soltanov 2mo ago|2 comments
35.
by soltanov 2mo ago
it is very beautifull idea and implementation. Open source available? It is good material for teaching how redis works.
36.
Ask HN: How to Protect Your Brand Before AI Copies It?
1 points by soltanov 2mo ago|2 comments
37.
Rethinking skills and prompts for GPT-6 Astra(developers.openai.com)
1 points by soltanov 2mo ago|0 comments
38.
by soltanov 2mo ago
Link: https://finance.yahoo.com/technology/ai/articles/ai-data-cen...
39.
Bounded Ratios for Lorentzian Polynomials(arxiv.org)
2 points by soltanov 2mo ago|1 comments
40.
by soltanov 2mo ago
Check also: https://arxiv.org/pdf/2609.05341
41.
Unsolved Problem by Fields Medalist Breached by Two High School Students(htx.com)
33 points by soltanov 2mo ago|59 comments
42.
AI Data Centers Are on Track to Fuel 'Explosive' Growth in Captive Insurance(finance.yahoo.com)
2 points by soltanov 2mo ago|2 comments
43.
PaperCut Attacker (Russian Linked) Uses AI Agents to Compromise 440 Instances(swapupdate.in)
2 points by soltanov 2mo ago|0 comments
44.
Breakfast is served autonomously now(cheffy.ai)
3 points by soltanov 2mo ago|3 comments
45.
by soltanov 2mo ago
Check also: https://www.instagram.com/reel/DZNcuI4hwHL/
46.
by soltanov 2mo ago
The interesting benchmark is not merely whether each agent receives a VM. It is whether a compromised task can cross capability, credential, network, or persistence boundaries.
47.
by soltanov 2mo ago
Separate model behavior from system security. Even an unreliable model can be deployed safely if capabilities, egress, credentials, and state transitions are enforced outside it.
48.
by soltanov 2mo ago
Does the protocol preserve its privacy benefit when payment, recovery, and anti-abuse metadata are considered together?
49.
by soltanov 2mo ago
This appears less like a classification-capability problem than an incentives and enforcement problem. What measurable abuse-response SLA should an ad platform provide?
50.
by soltanov 2mo ago
What evidence would make this a verified research result—independent replication, exact source scans, a deterministic decoding script, or all three?
51.
by soltanov 2mo ago
For any high-trust system, users need to know exactly what was verified, what was not, and which assumptions the verification depends on.
52.
by soltanov 2mo ago
Old platform limitations do not remove an application’s duty to minimize collection and offer clear controls
53.
by soltanov 2mo ago
Privacy claims should be tested with controlled egress checks, not inferred from a settings screen.
54.
by soltanov 2mo ago
Multi-agent tools need more than a visual dashboard
55.
Artemis: Google's new AI agent framework for mobile test automation(github.com)
2 points by soltanov 2mo ago|0 comments
56.
Maven Robotics wants to steal your robot deployment deal(techcrunch.com)
2 points by soltanov 2mo ago|0 comments
57.
Anthropic reveals rogue AI agents hate CAPTCHAs(techcrunch.com)
2 points by soltanov 2mo ago|0 comments
58.
Google Launches Meridian GeoX Globally(developers.google.com)
2 points by soltanov 2mo ago|0 comments
59.
by soltanov 2mo ago
Everyone shares their all kinds of data with ai companies. Privacy is dead anyway
60.
Redis built a cache that cuts LLM costs(redis.io)
1 points by soltanov 2mo ago|0 comments

More ›