Monthly Tech Bites #36 | U.S. Presidential Elections, AI Agents, Generative AI in Banking, and Notes on the Future and Security of AI

Hello there!

AI is moving faster than ever, bringing incredible opportunities – and serious questions.

Who’s in charge? How do we keep it ethical? Can we make it secure?

With business leaders like Elon Musk stepping into politics, the line between tech and power is blurring. Meanwhile, AI agents like Claude are becoming smarter and more independent, opening the door to amazing possibilities – but also new risks.

This week, let’s talk about the intersection of innovation, responsibility, and security in AI.

Because the future isn’t just about what AI can achieve. It’s also about how we shape it and ensure it’s used responsibly.

Here's what I've got for you today:

?? How will the US presidential election affect the future of AI?

?? Streaming 2030: The future of music consumption

?? Everyone’s excited about AI agents. But what are they, really?

?? Nest Bank’s N!Assistant: a game-changer in Polish banking

?? 5 essential tips to keep generative AI secure

Let’s dive right in!

How will the US presidential election affect the future of AI?

By now, everyone knows the results of the U.S. presidential election. Donald Trump is returning to the White House. Today, I want to look at what this means for the global economy and technology, especially AI.

During the campaign, Elon Musk strongly supported Trump. Musk, who owns the social platform X, SpaceX, and the AI company xAI, has been appointed co-leader of the new Department of Government Efficiency. This role gives him more influence over AI policy and government operations.

Michael Spencer , in his brilliant article published on AI Supremacy, argues that Trump’s win makes Musk “The Chief AI Officer of America.” He also points out that Musk’s platform X has been accused of algorithmic adjustments that boosted Trump’s campaign, raising questions about Musk’s political motives and his control over public conversation.

So, what does this mean for AI?

Musk’s company, xAI, is growing fast and challenging OpenAI and Anthropic. It recently raised $6 billion, valuing it at $50 billion, with major support from Middle Eastern investors. xAI is also doubling the power of its Colossus AI supercomputer, making Musk a key player in AI development.

This raises concerns about conflicts of interest, especially with foreign investments like Saudi Arabia’s Public Investment Fund. It also highlights the global stakes of AI leadership.

Spencer calls this a move toward “tycoon capitalism,” where billionaires like Musk shape politics and technology.

I’m not here to judge the U.S. election, but I agree with Spencer’s point –? business leaders behind powerful technologies like AI need to think critically about the impact of their work and act ethically and responsibly for the good of all of us.

Let’s hope Musk shares this perspective and the AI race remains ethical and fair.

Streaming 2030: The future of music consumption

Have you ever wondered how music streaming will evolve in the next decade? With 100,000 new tracks uploaded daily, it’s becoming harder for music to stand out. So, what’s next for artists and listeners?

Miquido is teaming up with Music Ally for an exciting event – Streaming 2030: The Future of Music Consumption. This session will explore key trends, including:

• AI-driven recommendations and how they’ll personalize music discovery for listeners.
• The rise of superfan tiers and how artists can build stronger, more engaged fan communities.
• How tipping models could revolutionize fan support for musicians.
• The latest advances in audio quality and what comes after hi-res and spatial audio.
• The growing trend of social music listening, making streaming more interactive.
• The potential impact of AI-generated music in future streaming catalogs.

The event will feature industry leaders like Raoul Chatterjee (Diverge), Pete Downton (Miquido), vickie nauman (CrossBorderWorks), and Linnea Sundberg (United Masters), with Joe Sparrow from Music Ally moderating.

Join us on 27 Nov to hear from the experts shaping the future of music streaming.

?? Register now and be part of the conversation!

Everyone’s excited about AI agents. But what are they, really?

I still remember a lively debate during one of our AI team meetings about agents: are they just tools designed to handle specific tasks, or are they more like autonomous systems capable of working on their own with minimal human input?

Most of the team leaned toward the latter, but over a year ago, we weren’t sure if such independence was truly achievable.

Today, the answer is much clearer. In his latest article, Ethan Mollick cuts through the confusion: an AI agent is an AI that acts independently to achieve goals set by the user.

And the best thing? Now, we’re starting to see real examples of these agents in action.

Claude: autonomous AI in action

Mollick uses Anthropic’s upgraded Claude 3.5 Sonnet model (in beta) as an example of how AI agents should work.

The upgraded Claude 3.5 Sonnet model is capable of interacting with tools that can manipulate a computer desktop environment.

Claude can handle complex tasks, like creating lesson plans or navigating games, all with very little input. It learns from feedback, adjusts its approach, and even builds its own tools to get the job done.

Of course, Claude isn’t perfect. It sometimes struggles with logic, depth, and personalization, which can lead to mistakes or generic results. Its independence can also make it hard to steer once it starts a task. But these are problems that will likely improve over time.

How to make AI agents work for you?

AI agents like Claude are transforming how we work, serve customers, and manage daily operations. And they’re arriving faster than most people think.

So for businesses, the big question isn’t if you should adopt AI agents – it’s how to make them work for you.?

Imagine an AI that operates across devices, tackles complex tasks for you or your customers, and adapts to your unique needs. How would that reshape your workflows, products, or even your strategy?

Nest Bank’s N!Assistant: a game-changer in Polish banking

Independent AI agents are the future, but for industries like banking, that future feels… a little more distant. The problem isn’t a lack of potential – it’s the weight of regulation and risk.

From my experience with banking clients, deploying fully autonomous systems would raise too many red flags. But that doesn’t mean banks are standing still.

Poland’s first AI banking assistant

Nest Bank S.A. is breaking new ground with N!Assistant, the first AI-driven banking assistant in Poland. Powered by GPT-4, it lets customers chat in natural language, analyze expenses, and even complete banking tasks using voice commands. For now, it’s available to a select group of users in their mobile app, but its capabilities are set to grow over time.

The big shift to true AI

At this point, most banking chatbots are decision trees. They rely on machine learning to detect user intent and trigger predefined responses. Nest Bank, however, has taken a step forward by deploying an LLM instead.

But was that a secure choice?

In my view – absolutely. The key lies in thoughtful implementation. The right solution architecture and securing non-deterministic ML models with deterministic control mechanisms enable achieving incredible results with LLMs.

Nest Bank is showing how LLMs can work even in strict industries like banking when used responsibly. Their approach is a great example for others and hopefully gets more organizations thinking about how to use AI in the right way.

5 tips to keep generative AI secure

Speaking of security and responsibility, I couldn’t miss the chance to share this with you. The article summarizing our recent event – AI Waves, focused on cybersecurity in the age of AI – is now live on our blog!

In the article, we’ve highlighted the 5 most important lessons shared by our brilliant speakers: Diana Kelley , Deepak Prabhakara , and ?ukasz Boruń .

In short, here they are:

1?? Encrypt and anonymize data at every stage to protect sensitive information.

2?? Don’t overtrust AI – human oversight is essential to catch errors.

3?? Secure your infrastructure – opt for enterprise-grade models, not free-tier options.

4?? Defend against prompt attacks with input validation and adversarial training.

5?? Vet third-party models to avoid hidden vulnerabilities and bias.

Make sure to read our article featuring insights from AI security experts to learn more.

Want to watch the full session?

Click here to access the recording ??.

That’s it for today!

If you want to keep up with the latest news on AI and be a part of the exciting world of technology, ride this innovation wave with me and my Miquido AI Crew by joining our newsletter! ??