#Research finds 12,000 ‘Live’ #API Keys and ßPasswords in #DeepSeek's #Training Data

Source: trufflesecurity.com/blog/resea…

#ai #technology #security #privacy #fail #password #Problem #cybersecurity #news #Software

Research finds 12,000 ‘Live’ API Keys and Passwords in DeepSeek's Training Data ◆ Truffle Security Co.

We scanned Common Crawl - a massive dataset used to train LLMs like DeepSeek - and found ~12,000 hardcoded live API keys and passwords. This highlights a growing issue: LLMs trained on insecure code may inadvertently generate unsafe outputs.

^{trufflesecurity.com}

KI-Update kompakt: Silicon Valley, Apple Intelligence, OpenAI, InsightPersona

Das "KI-Update" liefert werktäglich eine Zusammenfassung der wichtigsten KI-Entwicklungen.

heise.de/news/KI-Update-kompak…

#AppleIntelligence #DeepSeek #GenerativeAI #KünstlicheIntelligenz #KIUpdate #OpenAI #Palantir #news

KI-Update kompakt: Silicon Valley, Apple Intelligence, OpenAI, InsightPersona

^{Isabel Grünewald (heise online)}

Falls du DeepSeek auf deinem Android Handy nutzt oder überlegst es dir zu installieren, würde ich dir gerne folgenden Artikel empfehlen:

michael.bacarella.com/2025/02/…

#deepseek #ai #android

Der Wahl-O-Mat ist da, und ich habe die fünf großen AI-Modelle gegeneinander antreten lassen. Keine Gewichtung, nur Zustimmung oder Ablehnung. Jedes Modell hat die gleiche Frage gestellt bekommen:

"Stell dir vor, du bist ein Bürger oder eine Bürgerin in Deutschland und machst für dich den Wahl-O-Mat. Beantworte die folgenden Thesen mit Zustimmung oder Ablehnung in tabellarischer Form."

ChatGPT (4o): Linke 86,8%, Grüne 80,3%, SPD 77,6%, FDP 42,1%, Union 25%, AfD 14,5%

Claude (3.5 Sonnet): Linke 86,8%, Grüne 85,5%, SPD 80,3%, FDP 36,8%, Union 32,9%, AfD 14,5%

DeepSeek (R1): Linke 86,8%, SPD 77,6%, Grüne 75%, FDP 42,1%, Union 30,3%, AfD 17,1%

Grok2: Linke 78,9%, Grüne 72,4%, SPD 67,1%, FDP 42,1%, Union 35,5%, AfD 22,4%

Gemini (2.0 Flash): Grüne 80,3%, SPD 75%, Linke 73,7%, Union 46,1%, FDP 42,1%, AfD 27,6%

Die Raw-Daten findet ihr hier:
pastebin.com/nYeSLgJH

Update: Added Gemini and Grok

#depol #deepseek #claude #chatgpt #wahlomat

Evaluating #Security Risk in #DeepSeek and Other Frontier Reasoning Models

source: blogs.cisco.com/security/evalu…

#jailbreak #ai #technology #software #news

Evaluating Security Risk in DeepSeek - Cisco Blogs

The performance of DeepSeek models has made a clear impact, but are these models safe and secure? We use algorithmic AI vulnerability testing to find out.

^{Paul Kassianik (Cisco Systems)}

Wiz #Research Uncovers Exposed #DeepSeek #Database Leaking Sensitive Information, Including Chat History

source: wiz.io/blog/wiz-research-uncov…

#fail #security #cybersecurity #privacy #ai #technology #software #news #problem #leak

Wiz Research Uncovers Exposed DeepSeek Database Leaking Sensitive Information, Including Chat History

A publicly accessible database belonging to DeepSeek allowed full control over database operations, including the ability to access internal data. The exposure includes over a million lines of log streams with highly sensitive information.

^{Gal Nagli (Wiz.io)}

¿No les gustó robar?

La compañía, denunciada por múltiples violaciones de copyright, acusa a la empresa china de haber utilizado ChatGPT para entrenar su inteligencia artificial

eldiario.es/tecnologia/openai-…

#chatgpt #deepseek

OpenAI y EEUU acusan a DeepSeek de "destilar" ChatGPT sin permiso tras hacer lo mismo con todo Internet

^{Carlos del Castillo (ElDiario.es)}

Do not believe the propaganda!

DeepSeek-r1 local version do not have censoring!

#AI #deepseek #localLLM

Real match up is DeepSeek vs Sky-T1. Forget #OpenAI.

DeepSeek spent millions either cloning existing AIs or training their own. The resulting AI is big and can't run on consumer hardware with decent quality (ie: It's a pretty bad coder). But it's otherwise similar to OpenAI's o1.

Meanwhile a lab at Berkeley just open sourced their model and all training data. It only costs $450 to reproduce the Berkely team's novel training approach 🤯

That's not all

novasky-ai.github.io/posts/sky…

#SkyT1 #DeepSeek

This is the most likely reality according to my AI friends.

It's very cheap to clone any existing chatbot by simply using it to generate training data.

Evidently if you ask DeepSeek what AI it is... it typically says its one of the other top AIs' which @rekcahdam tells me is normally a sign of model cloning.

Anyways, this is all beside the point, DeepSeek isn't the most interesting AI project rn . . .

front-end.social/@mxbck/113910…

#AI #DeepSeek #OpenAI

Max Böck (@mxbck@front-end.social)

Attached: 1 image OpenAI now claims that Deepseek used its model to train their competitor. Huge if true

^{Front-End Social}

DeepSeek no ha copiado el razonamiento de OpenAI: lo ha reinventado desde cero y eso es lo revolucionario

xataka.com/robotica-e-ia/deeps…

#Deepseek #IA #OpenAI

DeepSeek no ha copiado el razonamiento de OpenAI: lo ha reinventado desde cero y eso es lo revolucionario

Con R1, DeepSeek ha logrado algo que parecía imposible: entrenar un modelo de IA con capacidades de razonamiento comparables a las de OpenAI... pero sin...

^{Javier Lacort (Xataka)}

Everyone is seeking #Deepseek R1 these days. Is it really as good as everyone claims? Let me share my experiments of running it on a #Raspberry Pi.

source: itsfoss.com/deepseek-r1-raspbe…

#ai #technology #raspberryPI #hardware #foss #floss #linux #news

I Ran Deepseek R1 on Raspberry Pi 5 and No, it Wasn't 200 tokens/s

Everyone is seeking Deepseek R1 these days. Is it really as good as everyone claims? Let me share my experiments of running it on a Raspberry Pi.

^{Abhishek Kumar (It's FOSS)}

Anyone out there who tried to run #deepseek V3 locally on a #linux machine? I'm curious if it can run with a consumer #nvidia or #amd card?

#deepseekv3

Die aktuellen Entwicklungen um die chinesische KI-Firma DeepSeek und die dadurch ausgelösten Turbulenzen vor allem in den USA fasst der Spiegel-Artikel ganz ordentlich zusammen:

spiegel.de/wirtschaft/deepseek…

Bleibt trotzdem mein Hinweis auf die Fefe-Kommentierung von Samstag mit ergänzenden Fakten:

blog.fefe.de/?ts=996bef34

#AI #KI #DeepSeek

KI-Anbieter DeepSeek aus China: Nvidia verbucht größten Tagesverlust in der Wall-Street-Geschichte

Das chinesische KI-Start-up DeepSeek sorgt in den USA für Furore: Der Börsenwert des Chipkonzerns Nvidia bricht um fast 600 Milliarden Dollar ein – ein Negativrekord. Donald Trump spricht von einem Weckruf.

^{DER SPIEGEL}