K, that's *really* all from me on #eWeekChat now. Promise!!

Until next time!!!

A10: My advice for #AI in #cybersecurity is start experimenting in ‘safe mode’ – e.g. non-prod systems like plan, dev, test, QA (cf. #DevSecOps); post-incident review; or prod-adjacent systems like #O11Y & #SIEM. Use findings to build policy & governance. Iterate to build trust.

#eWeekChat

Quick bonus question, no time for this today but it was in the running notes, so ...

Q10: Also about the future: Your best advice to help companies prepare for the future of AI and cybersecurity?

#eWeekChat (via @JamesMaguire on X)

Next month on #eWeekChat: 2025 #Predictions! Gonna be huge, says James. I am looking forward to it already!

That is all from me today from #eWeekChat but you can always join in by posting with the hashtag! Main action was over on X with the awesome @JamesMaguire but I will keep my eye on Threads, Mastodon, BlueSky, and LinkedIn to chat with anyone else who has Xited.

A9: Do not believe the #AI hype, make your own reality! Pundits already saying #AI is in the ‘trough of disillusionment’ but it is only just starting. Experiment on use cases, personas, outcomes to help you be more secure, safer, better governed, less vulnerable. Iterate to get value.

#eWeekChat

Q9: Final question: A last Big Thought about AI and cybersecurity—what else should managers/buyers/providers know about this emerging sector?

#eWeekChat (via @JamesMaguire on X)

A8: IMO the biggest future challenges for #AI in #cybersecurity are #trust, cost, #skills. Skills will come quickly – already are. Cost will persist, esp. driven by #IT #infrastructure & #energy cost. But trust will always be a major issue. Every new algo hallucinates differently.

#eWeekChat

Q8: Let’s look ahead: what enduring challenges will AI-based cybersecurity deployments continue to face in the future?

#eWeekChat (via @JamesMaguire on X)

A7: LOLWUT, no! 😂 #Hackers & malicious actors always find new penetration vectors and #IT is increasingly complex, so there will always be new threats. #AI is helping both sides, but hackers don’t care about legal or ethical boundaries. It will stay a constant game of ‘whack-a-mole’.

#eWeekChat

Q7: How do you sense that is AI changing the hacker community? That is, will AI ever enable companies to be fully safe from hacking?

#eWeekChat (via @JamesMaguire on X)

A6: I don’t think #AI will impact #SecOps #jobs in a macro sense – 3Vs keep growing, as do #cyber threats - but must adjust at a micro level. Resourcing is likely to be just as intense, but will need to include more algorithm development, ‘prompt engineering’, #data management, etc.

#eWeekChat

Q6: What about artificial intelligence and security staff? Will companies require less security staff due to AI’s support?

#eWeekChat (via @JamesMaguire on X)

A5: LOL, not sure we have empirically tested ‘known good patterns’ for #AI yet, let alone ‘best practices’! I’d say #DataOps & #MLOps for #data mgmt & algo auditability, validations to ensure provability & drive #trust are good starting places, but we need more ‘real world’ research.

#eWeekChat

Q5: What Best Practices advice would you give companies to more effectively deploy AI in their security infrastructure?

#eWeekChat (via @JamesMaguire on X)

A4: For #AI in #cybersecurity, need to ‘trust but verify’. #SecOps should adopt validation and verification techniques like #RAG, A-B Testing, Cross-Validation, supervised training, manual data checks, etc. At least until earning granular trust in how each algo processes each data set.

#eWeekChat

Q4: How do you recommend addressing these #AI related #cybersecurity challenges?

#eWeekChat (via @JamesMaguire on X)

A3. (cont) Also note #security issues are not just about #AI in #cybersecurity but also about cybersecurity for AI. Managing #penetration and #data leakage (cf. #DataOps), ensuring #audit and control for AI #algorithms (cf. #MLOps), filtering for malicious outputs, etc. 2-way street

#eWeekChat

A3: Many challenges for #AI in #cybersecurity, but IMO most critical is #trust. AI still hallucinates way too much, for #security pros that means too many false positives, and even missing a real threat. Cost, power, skills, goals are big speed bumps, but trust is a massive roadblock.

#eWeekChat

Q3: What are the biggest challenges facing #security and #AI? Cost, staff, training?

#eWeekChat (via @JamesMaguire on X)