Ghana's virtual assets policy, stablecoin deposit risk and securing enterprise AI

1) Ghana sets out a national policy position on virtual assets

What happened:

Ghana’s Nov 2025 Virtual Assets Policy—issued by the Bank of Ghana with the SEC and FIC—notes 3 M+ users and 100+ registered VASPs, confirms VAs aren’t legal tender, enforces the Travel Rule, and plans a new VA regulator and national literacy campaign.

Bank of Ghana

Why it matters:

A coherent, activity and riskbased framework in an EM market both legitimizes compliant providers and prepares for stablecoindriven flows. Pair this with Standard Chartered’s EM analysis (see #2) for deposit flight and policy coordination implications.

Standard Chartered Bank

2) DeFi’s mainstream turn: Standard Chartered’s forecasts

What’s new:

Standard Chartered Global Research sees DeFi challenging TradFi, forecasting $2 T in tokenized real-world assets by 2028—on par with stablecoins— and up to $1 T shifting from EM bank deposits to stablecoins, with risks clustered in certain markets.

CoinDesk

Why it matters:

Two near-term board questions stand out:

Liability risk: How vulnerable are local deposits and fees if stablecoin rails keep scaling? RWA strategy: How and where to tokenize assets without breaching custody, disclosure, or AML rules? (See Reuters and others for broader context.)

Reuters

3) CCAF launches AI in Financial Services 2030 global surveys

What’s new:

The Cambridge Centre for Alternative Finance kicked off a threepart global study (industry, regulators, AI vendors), backed by partners including the World Bank, BIS, IMF and WEF, to build a shared evidence base on AI adoption, impact and oversight in finance.

Cambridge Judge Business School

Why it matters:

Expect more grounded benchmarking of AI maturity (governance, explainability, agentic workflows) across jurisdictions—useful inputs for 2026 roadmapping and supervisory dialogues.

4) From “digital” to Ai-ternative finance

The take:

Jeremy Takle (Finextra) argues the winners won’t bolt AI onto legacy workflows; they’ll rebuild around intelligence architecture—unified, realtime data + context frameworks + agentic systems that shift banks from reactive service to proactive financial partner. The piece frames AInative as a system level redesign rather than a grabbag of use cases

Finextra Research

Why it matters:

I f your AI plan is a patchwork of pilots, your moat will erode. The agenda moves to operating model, data topology, retrieval context layers, tooling permissions, and continuous risk assurance.

5) Regulators as innovation enablers: Ireland’s non waiver sandbox

The research:

An SSRN paper on the Central Bank of Ireland’s inaugural Innovation Sandbox (Combatting Financial Crime) shows how a nonwaiver sandbox can deliver outcomes by systematically building social capital— via routines like relationship manager brokerage, engagement cycles, and explicit boundary setting— accelerating regulatory learning without legal carveouts.

SSRN

Why it matters:

For AI/DeFi pilots under tight regimes (e.g., EU AI Act context), structure and trust matter as much as exemptions. Design the process, not just the tech.

6) The silent exposure crisis in enterprise AI

Signal:

As teams wire LLMs into workflows, exposure vectors multiply—prompt injections that trigger tool misuse, quiet cross app data leakage, and overprivileged agent actions. Treat agents like superusers (least privilege, gated tools, immutable logs, human in the loop on high impact actions).

Swiss AI Institute Help Net Security

Why it matters:

The adoption curve is outpacing assurance. Make “AI security & compliance by design” as routine as CI/CD: red team your prompts/tools, classify data at ingress, and monitor model/tool telemetry like any other critical service.

Help Net Security