July 12

Event takeaway – KI und Sicherheit: „Balanceakt – Sicherheit schaffen, Freiheit bewahren“

Security

0  comments

My key takeaways

  • the base for intelligence is perception and knowledge
  • a spam filter is a kind of AI according to Prof. Dr. Maucher
  • AI doesn't require domain knowledge anymore, if there is enough data
  • there are 4 basic types of AI:
    • search and plan
    • machine learning
    • knowledge and inference
    • modelling of uncertainty
  • today there are approx 52 mio faked documents in the wild (EU)
  • 2019 : 1,7 mio images to scan for CSAM in the LKA BW
  • approx 300 CSAM suspicious incidents per week reported to the LKA
  • black box phenomenon: after some processing of date in a AI system, it is not reconstructable, which data exactly have been used
  • Prof. Dr. Poscher questions if our construct of privacy law is not feasible anymore
    • today the processing of any PII is a interference with fundamental rights
    • privacy laws should protect us regarding abstract risks that exceed general life risk
    • maybe we need to develop a more systemic, behavioristic understanding for AI systems and data processing within such, instead of regulating the processing of single data points

Env

  • Provided by Der Landesbeauftragte für den Datenschutz und die Informationsfreiheit Baden-Württemberg

  • Presenter:

    • Benjamin Bäßler, Abteilungsleiter beim LfDI

  • Speaker

    • Prof. Dr. Johannes Maucher (Leiter Institut für angewandte künstliche Intelligenz (IAAI), Hochschule der Medien Stuttgart)
    • Andreas Stenger (Präsident des LKA BW)
    • Hr. Hauke, Ikara Deep Vision Solutions (Kaiserslautern)
    • Hr. Hümmeler (BfDi)
    • Prof. Dr. Ralf Poscher (Direktor Öffentliches Recht beim Max-Planck-Institut für Kriminalität, Sicherheit und Recht, Freiburg)

additional links

User comments

Tags

AI, GDPR, privacy

You may also like

AI Cyber Security 2025: Key Threats and Essential Strategies for Your Business

Artificial Intelligence is transforming businesses and industries worldwide, but with this transformation comes unprecedented challenges. The recent launch of Cisco AI Defense, a specialized AI security solution, highlights the urgency of securing AI applications. According to Cisco, only 29% of enterprises feel fully prepared to detect and prevent unauthorized manipulations of AI systems. This low

Read More

SPQA: A Radical AI Architecture—and Why Security Must Adapt

Introduction Until recently, many developers laughed at the notion that Artificial Intelligence might write their code someday. Yet code assistants like Cursor, Bolt.new, and Copilot—already adopted by 51% of enterprises [1]—show how swiftly AI is reshaping software development. At the same time, spending on generative AI skyrocketed from $2.3 billion in 2023 to $13.8 billion

Read More
Leave a Comment