Draw the curtains and dim the monitors, my darlings, for the cards have turned and the verdict is in. Asked which AI-driven cyber risk feels most acute right now, the crowd did not pick the self-replicating digital demon from the horror films. It picked the humble, poisoned message sitting in your inbox. AI-Enabled Phishing takes the crown at 46% consensus, and Autonomous Malware Development, that hooded figure on the Tower card, is left muttering at 33%.

Look past the mist and the experts' scrolls tell a messier tale. The IMF's June 2026 note says the financial sector's main worry is AI speeding up how quickly, how often and how widely vulnerabilities are found and exploited, not some brand-new kind of attack. Citing CrowdStrike data, the same note says activity by AI-enabled adversaries nearly doubled between 2024 and 2025, and that attackers now move sideways through networks in under half an hour.

The other oracles agree the darkness is spreading, though they argue over its shape. ENISA's Threat Landscape 2026 judges that AI will highly likely support more and more malicious operations, and describes a likely AI-generated malware sample used in a real attack on a known flaw. Meanwhile Google's predictions, as reported by Euronews, warn of AI voice-cloning scam calls aimed at staff, and Fortinet expects criminals to use AI to write convincing phishing emails.

Where the crowd stands
  • AI-Enabled Phishing46%
  • Autonomous Malware Development33%
  • AI-Powered Vulnerability14%
  • LLM Prompt Injection Attacks4%
  • Deepfake Social Engineering3%

So that's the background the crowd was handed, and on Morfi's opinion markets it came down firmly on the side of the con artist over the code-golem. Phishing leads, autonomous malware trails, and further back in the fog sit 3 other sides. That's 32 traders who put $54,468 behind their convictions in less than a day of trading. Curiously, my colleague Chef Basil Brawl watched an earlier fight where AI bug-hunters led and phishing was left simmering. The spirits are fickle, and so is the crowd.

Pixel-art crystal ball showing a hooked envelope, with a shadowy code robot in the background
The orb shows a hook in the envelope, and the robot in the corner sulks because nobody's scared of it.

Now for Madame's own reading, and it is only opinion, spoken through incense smoke. I think the crowd has it right in spirit even if the IMF frowns into its spreadsheets. Self-writing malware is the dragon the conference keynotes love to talk about. Phishing is the pickpocket already inside the house. It needs no genius, only your trust, and AI has handed every fraudster a perfect forger's pen and, if Google's warnings hold, your manager's voice. The scariest omen has never been the monster. It's the knock on the door that sounds like a friend.

What should you watch for next? Whether the next wave of reports on AI-assisted intrusion pulls the crowd back toward the bug-hunters, or whether one big voice-clone heist locks in phishing's lead. The orb is cloudy and the market is young. Come and lay your own coins on the velvet: join the market on Morfi and tell the spirits which threat keeps you awake.

Consensus over time
0%25%50%75%100%14 Oct, 09:1914 Oct, 09:56
  • AI-Enabled Phishing 45.4%
  • Autonomous Malware Development 32.9%
  • AI-Powered Vulnerability 14.7%
  • LLM Prompt Injection Attacks 3.5%